Assignment4.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "9374e0d2-cfb7-479c-8aae-861ba6b092f0",
   "metadata": {},
   "outputs": [],
   "source": [
    "from pyspark.sql import SparkSession\n",
    "from pyspark.sql import functions as F\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "af71bec5-75cc-4f2a-8ce2-d89dfede84c9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('spark.stage.maxConsecutiveAttempts', '10'),\n",
       " ('spark.dynamicAllocation.minExecutors', '1'),\n",
       " ('spark.eventLog.enabled', 'true'),\n",
       " ('spark.submit.pyFiles',\n",
       "  '/root/.ivy2/jars/com.johnsnowlabs.nlp_spark-nlp_2.12-4.4.0.jar,/root/.ivy2/jars/graphframes_graphframes-0.8.2-spark3.1-s_2.12.jar,/root/.ivy2/jars/com.typesafe_config-1.4.2.jar,/root/.ivy2/jars/org.rocksdb_rocksdbjni-6.29.5.jar,/root/.ivy2/jars/com.amazonaws_aws-java-sdk-bundle-1.11.828.jar,/root/.ivy2/jars/com.github.universal-automata_liblevenshtein-3.0.0.jar,/root/.ivy2/jars/com.google.cloud_google-cloud-storage-2.16.0.jar,/root/.ivy2/jars/com.navigamez_greex-1.0.jar,/root/.ivy2/jars/com.johnsnowlabs.nlp_tensorflow-cpu_2.12-0.4.4.jar,/root/.ivy2/jars/it.unimi.dsi_fastutil-7.0.12.jar,/root/.ivy2/jars/org.projectlombok_lombok-1.16.8.jar,/root/.ivy2/jars/com.google.guava_guava-31.1-jre.jar,/root/.ivy2/jars/com.google.guava_failureaccess-1.0.1.jar,/root/.ivy2/jars/com.google.guava_listenablefuture-9999.0-empty-to-avoid-conflict-with-guava.jar,/root/.ivy2/jars/com.google.errorprone_error_prone_annotations-2.16.jar,/root/.ivy2/jars/com.google.j2objc_j2objc-annotations-1.3.jar,/root/.ivy2/jars/com.google.http-client_google-http-client-1.42.3.jar,/root/.ivy2/jars/io.opencensus_opencensus-contrib-http-util-0.31.1.jar,/root/.ivy2/jars/com.google.http-client_google-http-client-jackson2-1.42.3.jar,/root/.ivy2/jars/com.google.http-client_google-http-client-gson-1.42.3.jar,/root/.ivy2/jars/com.google.api-client_google-api-client-2.1.1.jar,/root/.ivy2/jars/commons-codec_commons-codec-1.15.jar,/root/.ivy2/jars/com.google.oauth-client_google-oauth-client-1.34.1.jar,/root/.ivy2/jars/com.google.http-client_google-http-client-apache-v2-1.42.3.jar,/root/.ivy2/jars/com.google.apis_google-api-services-storage-v1-rev20220705-2.0.0.jar,/root/.ivy2/jars/com.google.code.gson_gson-2.10.jar,/root/.ivy2/jars/com.google.cloud_google-cloud-core-2.9.0.jar,/root/.ivy2/jars/com.google.auto.value_auto-value-annotations-1.10.1.jar,/root/.ivy2/jars/com.google.cloud_google-cloud-core-http-2.9.0.jar,/root/.ivy2/jars/com.google.http-client_google-http-client-appengine-1.42.3.jar,/root/.ivy2/jars/com.google.api_gax-httpjson-0.105.1.jar,/root/.ivy2/jars/com.google.cloud_google-cloud-core-grpc-2.9.0.jar,/root/.ivy2/jars/io.grpc_grpc-core-1.51.0.jar,/root/.ivy2/jars/com.google.api_gax-2.20.1.jar,/root/.ivy2/jars/com.google.api_gax-grpc-2.20.1.jar,/root/.ivy2/jars/io.grpc_grpc-alts-1.51.0.jar,/root/.ivy2/jars/io.grpc_grpc-grpclb-1.51.0.jar,/root/.ivy2/jars/org.conscrypt_conscrypt-openjdk-uber-2.5.2.jar,/root/.ivy2/jars/io.grpc_grpc-protobuf-1.51.0.jar,/root/.ivy2/jars/com.google.auth_google-auth-library-credentials-1.13.0.jar,/root/.ivy2/jars/com.google.auth_google-auth-library-oauth2-http-1.13.0.jar,/root/.ivy2/jars/com.google.api_api-common-2.2.2.jar,/root/.ivy2/jars/javax.annotation_javax.annotation-api-1.3.2.jar,/root/.ivy2/jars/io.opencensus_opencensus-api-0.31.1.jar,/root/.ivy2/jars/io.grpc_grpc-context-1.51.0.jar,/root/.ivy2/jars/com.google.api.grpc_proto-google-iam-v1-1.6.22.jar,/root/.ivy2/jars/com.google.protobuf_protobuf-java-3.21.10.jar,/root/.ivy2/jars/com.google.protobuf_protobuf-java-util-3.21.10.jar,/root/.ivy2/jars/com.google.api.grpc_proto-google-common-protos-2.11.0.jar,/root/.ivy2/jars/org.threeten_threetenbp-1.6.4.jar,/root/.ivy2/jars/com.google.api.grpc_proto-google-cloud-storage-v2-2.16.0-alpha.jar,/root/.ivy2/jars/com.google.api.grpc_grpc-google-cloud-storage-v2-2.16.0-alpha.jar,/root/.ivy2/jars/com.google.api.grpc_gapic-google-cloud-storage-v2-2.16.0-alpha.jar,/root/.ivy2/jars/com.fasterxml.jackson.core_jackson-core-2.14.1.jar,/root/.ivy2/jars/com.google.code.findbugs_jsr305-3.0.2.jar,/root/.ivy2/jars/io.grpc_grpc-api-1.51.0.jar,/root/.ivy2/jars/io.grpc_grpc-auth-1.51.0.jar,/root/.ivy2/jars/io.grpc_grpc-stub-1.51.0.jar,/root/.ivy2/jars/org.checkerframework_checker-qual-3.28.0.jar,/root/.ivy2/jars/com.google.api.grpc_grpc-google-iam-v1-1.6.22.jar,/root/.ivy2/jars/io.grpc_grpc-protobuf-lite-1.51.0.jar,/root/.ivy2/jars/com.google.android_annotations-4.1.1.4.jar,/root/.ivy2/jars/org.codehaus.mojo_animal-sniffer-annotations-1.22.jar,/root/.ivy2/jars/io.grpc_grpc-netty-shaded-1.51.0.jar,/root/.ivy2/jars/io.perfmark_perfmark-api-0.26.0.jar,/root/.ivy2/jars/io.grpc_grpc-googleapis-1.51.0.jar,/root/.ivy2/jars/io.grpc_grpc-xds-1.51.0.jar,/root/.ivy2/jars/io.opencensus_opencensus-proto-0.2.0.jar,/root/.ivy2/jars/io.grpc_grpc-services-1.51.0.jar,/root/.ivy2/jars/com.google.re2j_re2j-1.6.jar,/root/.ivy2/jars/dk.brics.automaton_automaton-1.11-8.jar,/root/.ivy2/jars/org.slf4j_slf4j-api-1.7.16.jar'),\n",
       " ('spark.dataproc.sql.joinConditionReorder.enabled', 'true'),\n",
       " ('spark.cores.max', '16'),\n",
       " ('spark.sql.autoBroadcastJoinThreshold', '191m'),\n",
       " ('spark.org.apache.hadoop.yarn.server.webproxy.amfilter.AmIpFilter.param.PROXY_URI_BASES',\n",
       "  'http://hub-msca-bdp-dphub-students-test-ridhi-m:8088/proxy/application_1701103984606_0002'),\n",
       " ('spark.kryoserializer.buffer.max', '2000M'),\n",
       " ('spark.app.id', 'application_1701103984606_0002'),\n",
       " ('spark.serializer', 'org.apache.spark.serializer.KryoSerializer'),\n",
       " ('spark.dataproc.sql.local.rank.pushdown.enabled', 'true'),\n",
       " ('spark.driver.maxResultSize', '0'),\n",
       " ('spark.yarn.unmanagedAM.enabled', 'true'),\n",
       " ('spark.ui.filters',\n",
       "  'org.apache.hadoop.yarn.server.webproxy.amfilter.AmIpFilter'),\n",
       " ('spark.metrics.namespace',\n",
       "  'app_name:${spark.app.name}.app_id:${spark.app.id}'),\n",
       " ('spark.org.apache.hadoop.yarn.server.webproxy.amfilter.AmIpFilter.param.PROXY_HOSTS',\n",
       "  'hub-msca-bdp-dphub-students-test-ridhi-m'),\n",
       " ('spark.dataproc.sql.optimizer.leftsemijoin.conversion.enabled', 'true'),\n",
       " ('spark.hadoop.hive.execution.engine', 'mr'),\n",
       " ('spark.executor.id', 'driver'),\n",
       " ('spark.driver.host',\n",
       "  'hub-msca-bdp-dphub-students-test-ridhi-m.c.msca-bdp-student-ap.internal'),\n",
       " ('spark.hadoop.mapreduce.fileoutputcommitter.algorithm.version', '2'),\n",
       " ('spark.eventLog.dir',\n",
       "  'gs://dataproc-temp-us-central1-635155370842-uzamlpgc/6b0a4ee0-5210-49ee-a204-fbdcdb9c4ee3/spark-job-history'),\n",
       " ('spark.dynamicAllocation.maxExecutors', '10000'),\n",
       " ('spark.executor.memory', '32g'),\n",
       " ('spark.driver.appUIAddress',\n",
       "  'http://hub-msca-bdp-dphub-students-test-ridhi-m.c.msca-bdp-student-ap.internal:43563'),\n",
       " ('spark.yarn.dist.pyFiles',\n",
       "  'file:///root/.ivy2/jars/com.johnsnowlabs.nlp_spark-nlp_2.12-4.4.0.jar,file:///root/.ivy2/jars/graphframes_graphframes-0.8.2-spark3.1-s_2.12.jar,file:///root/.ivy2/jars/com.typesafe_config-1.4.2.jar,file:///root/.ivy2/jars/org.rocksdb_rocksdbjni-6.29.5.jar,file:///root/.ivy2/jars/com.amazonaws_aws-java-sdk-bundle-1.11.828.jar,file:///root/.ivy2/jars/com.github.universal-automata_liblevenshtein-3.0.0.jar,file:///root/.ivy2/jars/com.google.cloud_google-cloud-storage-2.16.0.jar,file:///root/.ivy2/jars/com.navigamez_greex-1.0.jar,file:///root/.ivy2/jars/com.johnsnowlabs.nlp_tensorflow-cpu_2.12-0.4.4.jar,file:///root/.ivy2/jars/it.unimi.dsi_fastutil-7.0.12.jar,file:///root/.ivy2/jars/org.projectlombok_lombok-1.16.8.jar,file:///root/.ivy2/jars/com.google.guava_guava-31.1-jre.jar,file:///root/.ivy2/jars/com.google.guava_failureaccess-1.0.1.jar,file:///root/.ivy2/jars/com.google.guava_listenablefuture-9999.0-empty-to-avoid-conflict-with-guava.jar,file:///root/.ivy2/jars/com.google.errorprone_error_prone_annotations-2.16.jar,file:///root/.ivy2/jars/com.google.j2objc_j2objc-annotations-1.3.jar,file:///root/.ivy2/jars/com.google.http-client_google-http-client-1.42.3.jar,file:///root/.ivy2/jars/io.opencensus_opencensus-contrib-http-util-0.31.1.jar,file:///root/.ivy2/jars/com.google.http-client_google-http-client-jackson2-1.42.3.jar,file:///root/.ivy2/jars/com.google.http-client_google-http-client-gson-1.42.3.jar,file:///root/.ivy2/jars/com.google.api-client_google-api-client-2.1.1.jar,file:///root/.ivy2/jars/commons-codec_commons-codec-1.15.jar,file:///root/.ivy2/jars/com.google.oauth-client_google-oauth-client-1.34.1.jar,file:///root/.ivy2/jars/com.google.http-client_google-http-client-apache-v2-1.42.3.jar,file:///root/.ivy2/jars/com.google.apis_google-api-services-storage-v1-rev20220705-2.0.0.jar,file:///root/.ivy2/jars/com.google.code.gson_gson-2.10.jar,file:///root/.ivy2/jars/com.google.cloud_google-cloud-core-2.9.0.jar,file:///root/.ivy2/jars/com.google.auto.value_auto-value-annotations-1.10.1.jar,file:///root/.ivy2/jars/com.google.cloud_google-cloud-core-http-2.9.0.jar,file:///root/.ivy2/jars/com.google.http-client_google-http-client-appengine-1.42.3.jar,file:///root/.ivy2/jars/com.google.api_gax-httpjson-0.105.1.jar,file:///root/.ivy2/jars/com.google.cloud_google-cloud-core-grpc-2.9.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-core-1.51.0.jar,file:///root/.ivy2/jars/com.google.api_gax-2.20.1.jar,file:///root/.ivy2/jars/com.google.api_gax-grpc-2.20.1.jar,file:///root/.ivy2/jars/io.grpc_grpc-alts-1.51.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-grpclb-1.51.0.jar,file:///root/.ivy2/jars/org.conscrypt_conscrypt-openjdk-uber-2.5.2.jar,file:///root/.ivy2/jars/io.grpc_grpc-protobuf-1.51.0.jar,file:///root/.ivy2/jars/com.google.auth_google-auth-library-credentials-1.13.0.jar,file:///root/.ivy2/jars/com.google.auth_google-auth-library-oauth2-http-1.13.0.jar,file:///root/.ivy2/jars/com.google.api_api-common-2.2.2.jar,file:///root/.ivy2/jars/javax.annotation_javax.annotation-api-1.3.2.jar,file:///root/.ivy2/jars/io.opencensus_opencensus-api-0.31.1.jar,file:///root/.ivy2/jars/io.grpc_grpc-context-1.51.0.jar,file:///root/.ivy2/jars/com.google.api.grpc_proto-google-iam-v1-1.6.22.jar,file:///root/.ivy2/jars/com.google.protobuf_protobuf-java-3.21.10.jar,file:///root/.ivy2/jars/com.google.protobuf_protobuf-java-util-3.21.10.jar,file:///root/.ivy2/jars/com.google.api.grpc_proto-google-common-protos-2.11.0.jar,file:///root/.ivy2/jars/org.threeten_threetenbp-1.6.4.jar,file:///root/.ivy2/jars/com.google.api.grpc_proto-google-cloud-storage-v2-2.16.0-alpha.jar,file:///root/.ivy2/jars/com.google.api.grpc_grpc-google-cloud-storage-v2-2.16.0-alpha.jar,file:///root/.ivy2/jars/com.google.api.grpc_gapic-google-cloud-storage-v2-2.16.0-alpha.jar,file:///root/.ivy2/jars/com.fasterxml.jackson.core_jackson-core-2.14.1.jar,file:///root/.ivy2/jars/com.google.code.findbugs_jsr305-3.0.2.jar,file:///root/.ivy2/jars/io.grpc_grpc-api-1.51.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-auth-1.51.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-stub-1.51.0.jar,file:///root/.ivy2/jars/org.checkerframework_checker-qual-3.28.0.jar,file:///root/.ivy2/jars/com.google.api.grpc_grpc-google-iam-v1-1.6.22.jar,file:///root/.ivy2/jars/io.grpc_grpc-protobuf-lite-1.51.0.jar,file:///root/.ivy2/jars/com.google.android_annotations-4.1.1.4.jar,file:///root/.ivy2/jars/org.codehaus.mojo_animal-sniffer-annotations-1.22.jar,file:///root/.ivy2/jars/io.grpc_grpc-netty-shaded-1.51.0.jar,file:///root/.ivy2/jars/io.perfmark_perfmark-api-0.26.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-googleapis-1.51.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-xds-1.51.0.jar,file:///root/.ivy2/jars/io.opencensus_opencensus-proto-0.2.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-services-1.51.0.jar,file:///root/.ivy2/jars/com.google.re2j_re2j-1.6.jar,file:///root/.ivy2/jars/dk.brics.automaton_automaton-1.11-8.jar,file:///root/.ivy2/jars/org.slf4j_slf4j-api-1.7.16.jar'),\n",
       " ('spark.yarn.am.attemptFailuresValidityInterval', '1h'),\n",
       " ('spark.app.name', 'Spark Updated Conf'),\n",
       " ('spark.sql.catalogImplementation', 'hive'),\n",
       " ('spark.executorEnv.OPENBLAS_NUM_THREADS', '1'),\n",
       " ('spark.yarn.secondary.jars',\n",
       "  'com.johnsnowlabs.nlp_spark-nlp_2.12-4.4.0.jar,graphframes_graphframes-0.8.2-spark3.1-s_2.12.jar,com.typesafe_config-1.4.2.jar,org.rocksdb_rocksdbjni-6.29.5.jar,com.amazonaws_aws-java-sdk-bundle-1.11.828.jar,com.github.universal-automata_liblevenshtein-3.0.0.jar,com.google.cloud_google-cloud-storage-2.16.0.jar,com.navigamez_greex-1.0.jar,com.johnsnowlabs.nlp_tensorflow-cpu_2.12-0.4.4.jar,it.unimi.dsi_fastutil-7.0.12.jar,org.projectlombok_lombok-1.16.8.jar,com.google.guava_guava-31.1-jre.jar,com.google.guava_failureaccess-1.0.1.jar,com.google.guava_listenablefuture-9999.0-empty-to-avoid-conflict-with-guava.jar,com.google.errorprone_error_prone_annotations-2.16.jar,com.google.j2objc_j2objc-annotations-1.3.jar,com.google.http-client_google-http-client-1.42.3.jar,io.opencensus_opencensus-contrib-http-util-0.31.1.jar,com.google.http-client_google-http-client-jackson2-1.42.3.jar,com.google.http-client_google-http-client-gson-1.42.3.jar,com.google.api-client_google-api-client-2.1.1.jar,commons-codec_commons-codec-1.15.jar,com.google.oauth-client_google-oauth-client-1.34.1.jar,com.google.http-client_google-http-client-apache-v2-1.42.3.jar,com.google.apis_google-api-services-storage-v1-rev20220705-2.0.0.jar,com.google.code.gson_gson-2.10.jar,com.google.cloud_google-cloud-core-2.9.0.jar,com.google.auto.value_auto-value-annotations-1.10.1.jar,com.google.cloud_google-cloud-core-http-2.9.0.jar,com.google.http-client_google-http-client-appengine-1.42.3.jar,com.google.api_gax-httpjson-0.105.1.jar,com.google.cloud_google-cloud-core-grpc-2.9.0.jar,io.grpc_grpc-core-1.51.0.jar,com.google.api_gax-2.20.1.jar,com.google.api_gax-grpc-2.20.1.jar,io.grpc_grpc-alts-1.51.0.jar,io.grpc_grpc-grpclb-1.51.0.jar,org.conscrypt_conscrypt-openjdk-uber-2.5.2.jar,io.grpc_grpc-protobuf-1.51.0.jar,com.google.auth_google-auth-library-credentials-1.13.0.jar,com.google.auth_google-auth-library-oauth2-http-1.13.0.jar,com.google.api_api-common-2.2.2.jar,javax.annotation_javax.annotation-api-1.3.2.jar,io.opencensus_opencensus-api-0.31.1.jar,io.grpc_grpc-context-1.51.0.jar,com.google.api.grpc_proto-google-iam-v1-1.6.22.jar,com.google.protobuf_protobuf-java-3.21.10.jar,com.google.protobuf_protobuf-java-util-3.21.10.jar,com.google.api.grpc_proto-google-common-protos-2.11.0.jar,org.threeten_threetenbp-1.6.4.jar,com.google.api.grpc_proto-google-cloud-storage-v2-2.16.0-alpha.jar,com.google.api.grpc_grpc-google-cloud-storage-v2-2.16.0-alpha.jar,com.google.api.grpc_gapic-google-cloud-storage-v2-2.16.0-alpha.jar,com.fasterxml.jackson.core_jackson-core-2.14.1.jar,com.google.code.findbugs_jsr305-3.0.2.jar,io.grpc_grpc-api-1.51.0.jar,io.grpc_grpc-auth-1.51.0.jar,io.grpc_grpc-stub-1.51.0.jar,org.checkerframework_checker-qual-3.28.0.jar,com.google.api.grpc_grpc-google-iam-v1-1.6.22.jar,io.grpc_grpc-protobuf-lite-1.51.0.jar,com.google.android_annotations-4.1.1.4.jar,org.codehaus.mojo_animal-sniffer-annotations-1.22.jar,io.grpc_grpc-netty-shaded-1.51.0.jar,io.perfmark_perfmark-api-0.26.0.jar,io.grpc_grpc-googleapis-1.51.0.jar,io.grpc_grpc-xds-1.51.0.jar,io.opencensus_opencensus-proto-0.2.0.jar,io.grpc_grpc-services-1.51.0.jar,com.google.re2j_re2j-1.6.jar,dk.brics.automaton_automaton-1.11-8.jar,org.slf4j_slf4j-api-1.7.16.jar'),\n",
       " ('spark.repl.local.jars',\n",
       "  'file:///root/.ivy2/jars/com.johnsnowlabs.nlp_spark-nlp_2.12-4.4.0.jar,file:///root/.ivy2/jars/graphframes_graphframes-0.8.2-spark3.1-s_2.12.jar,file:///root/.ivy2/jars/com.typesafe_config-1.4.2.jar,file:///root/.ivy2/jars/org.rocksdb_rocksdbjni-6.29.5.jar,file:///root/.ivy2/jars/com.amazonaws_aws-java-sdk-bundle-1.11.828.jar,file:///root/.ivy2/jars/com.github.universal-automata_liblevenshtein-3.0.0.jar,file:///root/.ivy2/jars/com.google.cloud_google-cloud-storage-2.16.0.jar,file:///root/.ivy2/jars/com.navigamez_greex-1.0.jar,file:///root/.ivy2/jars/com.johnsnowlabs.nlp_tensorflow-cpu_2.12-0.4.4.jar,file:///root/.ivy2/jars/it.unimi.dsi_fastutil-7.0.12.jar,file:///root/.ivy2/jars/org.projectlombok_lombok-1.16.8.jar,file:///root/.ivy2/jars/com.google.guava_guava-31.1-jre.jar,file:///root/.ivy2/jars/com.google.guava_failureaccess-1.0.1.jar,file:///root/.ivy2/jars/com.google.guava_listenablefuture-9999.0-empty-to-avoid-conflict-with-guava.jar,file:///root/.ivy2/jars/com.google.errorprone_error_prone_annotations-2.16.jar,file:///root/.ivy2/jars/com.google.j2objc_j2objc-annotations-1.3.jar,file:///root/.ivy2/jars/com.google.http-client_google-http-client-1.42.3.jar,file:///root/.ivy2/jars/io.opencensus_opencensus-contrib-http-util-0.31.1.jar,file:///root/.ivy2/jars/com.google.http-client_google-http-client-jackson2-1.42.3.jar,file:///root/.ivy2/jars/com.google.http-client_google-http-client-gson-1.42.3.jar,file:///root/.ivy2/jars/com.google.api-client_google-api-client-2.1.1.jar,file:///root/.ivy2/jars/commons-codec_commons-codec-1.15.jar,file:///root/.ivy2/jars/com.google.oauth-client_google-oauth-client-1.34.1.jar,file:///root/.ivy2/jars/com.google.http-client_google-http-client-apache-v2-1.42.3.jar,file:///root/.ivy2/jars/com.google.apis_google-api-services-storage-v1-rev20220705-2.0.0.jar,file:///root/.ivy2/jars/com.google.code.gson_gson-2.10.jar,file:///root/.ivy2/jars/com.google.cloud_google-cloud-core-2.9.0.jar,file:///root/.ivy2/jars/com.google.auto.value_auto-value-annotations-1.10.1.jar,file:///root/.ivy2/jars/com.google.cloud_google-cloud-core-http-2.9.0.jar,file:///root/.ivy2/jars/com.google.http-client_google-http-client-appengine-1.42.3.jar,file:///root/.ivy2/jars/com.google.api_gax-httpjson-0.105.1.jar,file:///root/.ivy2/jars/com.google.cloud_google-cloud-core-grpc-2.9.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-core-1.51.0.jar,file:///root/.ivy2/jars/com.google.api_gax-2.20.1.jar,file:///root/.ivy2/jars/com.google.api_gax-grpc-2.20.1.jar,file:///root/.ivy2/jars/io.grpc_grpc-alts-1.51.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-grpclb-1.51.0.jar,file:///root/.ivy2/jars/org.conscrypt_conscrypt-openjdk-uber-2.5.2.jar,file:///root/.ivy2/jars/io.grpc_grpc-protobuf-1.51.0.jar,file:///root/.ivy2/jars/com.google.auth_google-auth-library-credentials-1.13.0.jar,file:///root/.ivy2/jars/com.google.auth_google-auth-library-oauth2-http-1.13.0.jar,file:///root/.ivy2/jars/com.google.api_api-common-2.2.2.jar,file:///root/.ivy2/jars/javax.annotation_javax.annotation-api-1.3.2.jar,file:///root/.ivy2/jars/io.opencensus_opencensus-api-0.31.1.jar,file:///root/.ivy2/jars/io.grpc_grpc-context-1.51.0.jar,file:///root/.ivy2/jars/com.google.api.grpc_proto-google-iam-v1-1.6.22.jar,file:///root/.ivy2/jars/com.google.protobuf_protobuf-java-3.21.10.jar,file:///root/.ivy2/jars/com.google.protobuf_protobuf-java-util-3.21.10.jar,file:///root/.ivy2/jars/com.google.api.grpc_proto-google-common-protos-2.11.0.jar,file:///root/.ivy2/jars/org.threeten_threetenbp-1.6.4.jar,file:///root/.ivy2/jars/com.google.api.grpc_proto-google-cloud-storage-v2-2.16.0-alpha.jar,file:///root/.ivy2/jars/com.google.api.grpc_grpc-google-cloud-storage-v2-2.16.0-alpha.jar,file:///root/.ivy2/jars/com.google.api.grpc_gapic-google-cloud-storage-v2-2.16.0-alpha.jar,file:///root/.ivy2/jars/com.fasterxml.jackson.core_jackson-core-2.14.1.jar,file:///root/.ivy2/jars/com.google.code.findbugs_jsr305-3.0.2.jar,file:///root/.ivy2/jars/io.grpc_grpc-api-1.51.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-auth-1.51.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-stub-1.51.0.jar,file:///root/.ivy2/jars/org.checkerframework_checker-qual-3.28.0.jar,file:///root/.ivy2/jars/com.google.api.grpc_grpc-google-iam-v1-1.6.22.jar,file:///root/.ivy2/jars/io.grpc_grpc-protobuf-lite-1.51.0.jar,file:///root/.ivy2/jars/com.google.android_annotations-4.1.1.4.jar,file:///root/.ivy2/jars/org.codehaus.mojo_animal-sniffer-annotations-1.22.jar,file:///root/.ivy2/jars/io.grpc_grpc-netty-shaded-1.51.0.jar,file:///root/.ivy2/jars/io.perfmark_perfmark-api-0.26.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-googleapis-1.51.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-xds-1.51.0.jar,file:///root/.ivy2/jars/io.opencensus_opencensus-proto-0.2.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-services-1.51.0.jar,file:///root/.ivy2/jars/com.google.re2j_re2j-1.6.jar,file:///root/.ivy2/jars/dk.brics.automaton_automaton-1.11-8.jar,file:///root/.ivy2/jars/org.slf4j_slf4j-api-1.7.16.jar'),\n",
       " ('spark.sql.cbo.enabled', 'true'),\n",
       " ('spark.executorEnv.PYTHONPATH',\n",
       "  '/usr/lib/spark/python/lib/py4j-0.10.9-src.zip:/usr/lib/spark/python/:<CPS>{{PWD}}/pyspark.zip<CPS>{{PWD}}/py4j-0.10.9-src.zip<CPS>{{PWD}}/com.johnsnowlabs.nlp_spark-nlp_2.12-4.4.0.jar<CPS>{{PWD}}/graphframes_graphframes-0.8.2-spark3.1-s_2.12.jar<CPS>{{PWD}}/com.typesafe_config-1.4.2.jar<CPS>{{PWD}}/org.rocksdb_rocksdbjni-6.29.5.jar<CPS>{{PWD}}/com.amazonaws_aws-java-sdk-bundle-1.11.828.jar<CPS>{{PWD}}/com.github.universal-automata_liblevenshtein-3.0.0.jar<CPS>{{PWD}}/com.google.cloud_google-cloud-storage-2.16.0.jar<CPS>{{PWD}}/com.navigamez_greex-1.0.jar<CPS>{{PWD}}/com.johnsnowlabs.nlp_tensorflow-cpu_2.12-0.4.4.jar<CPS>{{PWD}}/it.unimi.dsi_fastutil-7.0.12.jar<CPS>{{PWD}}/org.projectlombok_lombok-1.16.8.jar<CPS>{{PWD}}/com.google.guava_guava-31.1-jre.jar<CPS>{{PWD}}/com.google.guava_failureaccess-1.0.1.jar<CPS>{{PWD}}/com.google.guava_listenablefuture-9999.0-empty-to-avoid-conflict-with-guava.jar<CPS>{{PWD}}/com.google.errorprone_error_prone_annotations-2.16.jar<CPS>{{PWD}}/com.google.j2objc_j2objc-annotations-1.3.jar<CPS>{{PWD}}/com.google.http-client_google-http-client-1.42.3.jar<CPS>{{PWD}}/io.opencensus_opencensus-contrib-http-util-0.31.1.jar<CPS>{{PWD}}/com.google.http-client_google-http-client-jackson2-1.42.3.jar<CPS>{{PWD}}/com.google.http-client_google-http-client-gson-1.42.3.jar<CPS>{{PWD}}/com.google.api-client_google-api-client-2.1.1.jar<CPS>{{PWD}}/commons-codec_commons-codec-1.15.jar<CPS>{{PWD}}/com.google.oauth-client_google-oauth-client-1.34.1.jar<CPS>{{PWD}}/com.google.http-client_google-http-client-apache-v2-1.42.3.jar<CPS>{{PWD}}/com.google.apis_google-api-services-storage-v1-rev20220705-2.0.0.jar<CPS>{{PWD}}/com.google.code.gson_gson-2.10.jar<CPS>{{PWD}}/com.google.cloud_google-cloud-core-2.9.0.jar<CPS>{{PWD}}/com.google.auto.value_auto-value-annotations-1.10.1.jar<CPS>{{PWD}}/com.google.cloud_google-cloud-core-http-2.9.0.jar<CPS>{{PWD}}/com.google.http-client_google-http-client-appengine-1.42.3.jar<CPS>{{PWD}}/com.google.api_gax-httpjson-0.105.1.jar<CPS>{{PWD}}/com.google.cloud_google-cloud-core-grpc-2.9.0.jar<CPS>{{PWD}}/io.grpc_grpc-core-1.51.0.jar<CPS>{{PWD}}/com.google.api_gax-2.20.1.jar<CPS>{{PWD}}/com.google.api_gax-grpc-2.20.1.jar<CPS>{{PWD}}/io.grpc_grpc-alts-1.51.0.jar<CPS>{{PWD}}/io.grpc_grpc-grpclb-1.51.0.jar<CPS>{{PWD}}/org.conscrypt_conscrypt-openjdk-uber-2.5.2.jar<CPS>{{PWD}}/io.grpc_grpc-protobuf-1.51.0.jar<CPS>{{PWD}}/com.google.auth_google-auth-library-credentials-1.13.0.jar<CPS>{{PWD}}/com.google.auth_google-auth-library-oauth2-http-1.13.0.jar<CPS>{{PWD}}/com.google.api_api-common-2.2.2.jar<CPS>{{PWD}}/javax.annotation_javax.annotation-api-1.3.2.jar<CPS>{{PWD}}/io.opencensus_opencensus-api-0.31.1.jar<CPS>{{PWD}}/io.grpc_grpc-context-1.51.0.jar<CPS>{{PWD}}/com.google.api.grpc_proto-google-iam-v1-1.6.22.jar<CPS>{{PWD}}/com.google.protobuf_protobuf-java-3.21.10.jar<CPS>{{PWD}}/com.google.protobuf_protobuf-java-util-3.21.10.jar<CPS>{{PWD}}/com.google.api.grpc_proto-google-common-protos-2.11.0.jar<CPS>{{PWD}}/org.threeten_threetenbp-1.6.4.jar<CPS>{{PWD}}/com.google.api.grpc_proto-google-cloud-storage-v2-2.16.0-alpha.jar<CPS>{{PWD}}/com.google.api.grpc_grpc-google-cloud-storage-v2-2.16.0-alpha.jar<CPS>{{PWD}}/com.google.api.grpc_gapic-google-cloud-storage-v2-2.16.0-alpha.jar<CPS>{{PWD}}/com.fasterxml.jackson.core_jackson-core-2.14.1.jar<CPS>{{PWD}}/com.google.code.findbugs_jsr305-3.0.2.jar<CPS>{{PWD}}/io.grpc_grpc-api-1.51.0.jar<CPS>{{PWD}}/io.grpc_grpc-auth-1.51.0.jar<CPS>{{PWD}}/io.grpc_grpc-stub-1.51.0.jar<CPS>{{PWD}}/org.checkerframework_checker-qual-3.28.0.jar<CPS>{{PWD}}/com.google.api.grpc_grpc-google-iam-v1-1.6.22.jar<CPS>{{PWD}}/io.grpc_grpc-protobuf-lite-1.51.0.jar<CPS>{{PWD}}/com.google.android_annotations-4.1.1.4.jar<CPS>{{PWD}}/org.codehaus.mojo_animal-sniffer-annotations-1.22.jar<CPS>{{PWD}}/io.grpc_grpc-netty-shaded-1.51.0.jar<CPS>{{PWD}}/io.perfmark_perfmark-api-0.26.0.jar<CPS>{{PWD}}/io.grpc_grpc-googleapis-1.51.0.jar<CPS>{{PWD}}/io.grpc_grpc-xds-1.51.0.jar<CPS>{{PWD}}/io.opencensus_opencensus-proto-0.2.0.jar<CPS>{{PWD}}/io.grpc_grpc-services-1.51.0.jar<CPS>{{PWD}}/com.google.re2j_re2j-1.6.jar<CPS>{{PWD}}/dk.brics.automaton_automaton-1.11-8.jar<CPS>{{PWD}}/org.slf4j_slf4j-api-1.7.16.jar'),\n",
       " ('spark.yarn.dist.jars',\n",
       "  'file:///root/.ivy2/jars/com.johnsnowlabs.nlp_spark-nlp_2.12-4.4.0.jar,file:///root/.ivy2/jars/graphframes_graphframes-0.8.2-spark3.1-s_2.12.jar,file:///root/.ivy2/jars/com.typesafe_config-1.4.2.jar,file:///root/.ivy2/jars/org.rocksdb_rocksdbjni-6.29.5.jar,file:///root/.ivy2/jars/com.amazonaws_aws-java-sdk-bundle-1.11.828.jar,file:///root/.ivy2/jars/com.github.universal-automata_liblevenshtein-3.0.0.jar,file:///root/.ivy2/jars/com.google.cloud_google-cloud-storage-2.16.0.jar,file:///root/.ivy2/jars/com.navigamez_greex-1.0.jar,file:///root/.ivy2/jars/com.johnsnowlabs.nlp_tensorflow-cpu_2.12-0.4.4.jar,file:///root/.ivy2/jars/it.unimi.dsi_fastutil-7.0.12.jar,file:///root/.ivy2/jars/org.projectlombok_lombok-1.16.8.jar,file:///root/.ivy2/jars/com.google.guava_guava-31.1-jre.jar,file:///root/.ivy2/jars/com.google.guava_failureaccess-1.0.1.jar,file:///root/.ivy2/jars/com.google.guava_listenablefuture-9999.0-empty-to-avoid-conflict-with-guava.jar,file:///root/.ivy2/jars/com.google.errorprone_error_prone_annotations-2.16.jar,file:///root/.ivy2/jars/com.google.j2objc_j2objc-annotations-1.3.jar,file:///root/.ivy2/jars/com.google.http-client_google-http-client-1.42.3.jar,file:///root/.ivy2/jars/io.opencensus_opencensus-contrib-http-util-0.31.1.jar,file:///root/.ivy2/jars/com.google.http-client_google-http-client-jackson2-1.42.3.jar,file:///root/.ivy2/jars/com.google.http-client_google-http-client-gson-1.42.3.jar,file:///root/.ivy2/jars/com.google.api-client_google-api-client-2.1.1.jar,file:///root/.ivy2/jars/commons-codec_commons-codec-1.15.jar,file:///root/.ivy2/jars/com.google.oauth-client_google-oauth-client-1.34.1.jar,file:///root/.ivy2/jars/com.google.http-client_google-http-client-apache-v2-1.42.3.jar,file:///root/.ivy2/jars/com.google.apis_google-api-services-storage-v1-rev20220705-2.0.0.jar,file:///root/.ivy2/jars/com.google.code.gson_gson-2.10.jar,file:///root/.ivy2/jars/com.google.cloud_google-cloud-core-2.9.0.jar,file:///root/.ivy2/jars/com.google.auto.value_auto-value-annotations-1.10.1.jar,file:///root/.ivy2/jars/com.google.cloud_google-cloud-core-http-2.9.0.jar,file:///root/.ivy2/jars/com.google.http-client_google-http-client-appengine-1.42.3.jar,file:///root/.ivy2/jars/com.google.api_gax-httpjson-0.105.1.jar,file:///root/.ivy2/jars/com.google.cloud_google-cloud-core-grpc-2.9.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-core-1.51.0.jar,file:///root/.ivy2/jars/com.google.api_gax-2.20.1.jar,file:///root/.ivy2/jars/com.google.api_gax-grpc-2.20.1.jar,file:///root/.ivy2/jars/io.grpc_grpc-alts-1.51.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-grpclb-1.51.0.jar,file:///root/.ivy2/jars/org.conscrypt_conscrypt-openjdk-uber-2.5.2.jar,file:///root/.ivy2/jars/io.grpc_grpc-protobuf-1.51.0.jar,file:///root/.ivy2/jars/com.google.auth_google-auth-library-credentials-1.13.0.jar,file:///root/.ivy2/jars/com.google.auth_google-auth-library-oauth2-http-1.13.0.jar,file:///root/.ivy2/jars/com.google.api_api-common-2.2.2.jar,file:///root/.ivy2/jars/javax.annotation_javax.annotation-api-1.3.2.jar,file:///root/.ivy2/jars/io.opencensus_opencensus-api-0.31.1.jar,file:///root/.ivy2/jars/io.grpc_grpc-context-1.51.0.jar,file:///root/.ivy2/jars/com.google.api.grpc_proto-google-iam-v1-1.6.22.jar,file:///root/.ivy2/jars/com.google.protobuf_protobuf-java-3.21.10.jar,file:///root/.ivy2/jars/com.google.protobuf_protobuf-java-util-3.21.10.jar,file:///root/.ivy2/jars/com.google.api.grpc_proto-google-common-protos-2.11.0.jar,file:///root/.ivy2/jars/org.threeten_threetenbp-1.6.4.jar,file:///root/.ivy2/jars/com.google.api.grpc_proto-google-cloud-storage-v2-2.16.0-alpha.jar,file:///root/.ivy2/jars/com.google.api.grpc_grpc-google-cloud-storage-v2-2.16.0-alpha.jar,file:///root/.ivy2/jars/com.google.api.grpc_gapic-google-cloud-storage-v2-2.16.0-alpha.jar,file:///root/.ivy2/jars/com.fasterxml.jackson.core_jackson-core-2.14.1.jar,file:///root/.ivy2/jars/com.google.code.findbugs_jsr305-3.0.2.jar,file:///root/.ivy2/jars/io.grpc_grpc-api-1.51.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-auth-1.51.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-stub-1.51.0.jar,file:///root/.ivy2/jars/org.checkerframework_checker-qual-3.28.0.jar,file:///root/.ivy2/jars/com.google.api.grpc_grpc-google-iam-v1-1.6.22.jar,file:///root/.ivy2/jars/io.grpc_grpc-protobuf-lite-1.51.0.jar,file:///root/.ivy2/jars/com.google.android_annotations-4.1.1.4.jar,file:///root/.ivy2/jars/org.codehaus.mojo_animal-sniffer-annotations-1.22.jar,file:///root/.ivy2/jars/io.grpc_grpc-netty-shaded-1.51.0.jar,file:///root/.ivy2/jars/io.perfmark_perfmark-api-0.26.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-googleapis-1.51.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-xds-1.51.0.jar,file:///root/.ivy2/jars/io.opencensus_opencensus-proto-0.2.0.jar,file:///root/.ivy2/jars/io.grpc_grpc-services-1.51.0.jar,file:///root/.ivy2/jars/com.google.re2j_re2j-1.6.jar,file:///root/.ivy2/jars/dk.brics.automaton_automaton-1.11-8.jar,file:///root/.ivy2/jars/org.slf4j_slf4j-api-1.7.16.jar'),\n",
       " ('spark.dataproc.sql.parquet.enableFooterCache', 'true'),\n",
       " ('spark.driver.memory', '32g'),\n",
       " ('spark.sql.warehouse.dir', 'file:/spark-warehouse'),\n",
       " ('spark.yarn.executor.failuresValidityInterval', '1h'),\n",
       " ('spark.dataproc.metrics.listener.metrics.collector.hostname',\n",
       "  'hub-msca-bdp-dphub-students-test-ridhi-m'),\n",
       " ('spark.yarn.am.memory', '640m'),\n",
       " ('spark.ui.proxyBase', '/proxy/application_1701103984606_0002'),\n",
       " ('spark.jars.packages',\n",
       "  'com.johnsnowlabs.nlp:spark-nlp_2.12:4.4.0,graphframes:graphframes:0.8.2-spark3.1-s_2.12'),\n",
       " ('spark.executor.instances', '2'),\n",
       " ('spark.dataproc.listeners',\n",
       "  'com.google.cloud.spark.performance.DataprocMetricsListener'),\n",
       " ('spark.driver.port', '43589'),\n",
       " ('spark.serializer.objectStreamReset', '100'),\n",
       " ('spark.executor.cores', '16'),\n",
       " ('spark.submit.deployMode', 'client'),\n",
       " ('spark.history.fs.logDirectory',\n",
       "  'gs://dataproc-temp-us-central1-635155370842-uzamlpgc/6b0a4ee0-5210-49ee-a204-fbdcdb9c4ee3/spark-job-history'),\n",
       " ('spark.sql.cbo.joinReorder.enabled', 'true'),\n",
       " ('spark.shuffle.service.enabled', 'true'),\n",
       " ('spark.scheduler.mode', 'FAIR'),\n",
       " ('spark.sql.adaptive.enabled', 'true'),\n",
       " ('spark.yarn.jars', 'local:/usr/lib/spark/jars/*'),\n",
       " ('spark.scheduler.minRegisteredResourcesRatio', '0.0'),\n",
       " ('spark.master', 'yarn'),\n",
       " ('spark.ui.port', '0'),\n",
       " ('spark.rpc.message.maxSize', '512'),\n",
       " ('spark.rdd.compress', 'True'),\n",
       " ('spark.task.maxFailures', '10'),\n",
       " ('spark.yarn.isPython', 'true'),\n",
       " ('spark.dynamicAllocation.enabled', 'true'),\n",
       " ('spark.yarn.historyServer.address',\n",
       "  'hub-msca-bdp-dphub-students-test-ridhi-m:18080'),\n",
       " ('spark.ui.showConsoleProgress', 'true'),\n",
       " ('spark.app.startTime', '1701112785247')]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#create Spark session\n",
    "spark = SparkSession.builder.appName('movieTags').config(\"spark.broadcast.blockSize\", \"64m\").getOrCreate()\n",
    "\n",
    "conf = spark.sparkContext._conf.setAll([('spark.executor.memory', '32g'), ('spark.app.name', 'Spark Updated Conf'),\\\n",
    "                                        ('spark.executor.cores', '16'), ('spark.cores.max', '16'), ('spark.driver.memory','32g')])\n",
    "spark.sparkContext.setLogLevel('ERROR')\n",
    "spark.sparkContext.getConf().getAll()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "f4e02d14-cd1e-4166-9383-ddb8e52f4531",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'http://hub-msca-bdp-dphub-students-test-ridhi-m.c.msca-bdp-student-ap.internal:43563'"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "spark.sparkContext.uiWebUrl"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "881cbe87-07f2-44d1-a660-18c853973218",
   "metadata": {},
   "source": [
    "## 1) Data Exploration\n",
    "\n",
    "***a. Read the movie synopsis data into Spark***"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "04ddb818-d813-4089-a723-ed598cc187a9",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                                \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- imdb_id: string (nullable = true)\n",
      " |-- title: string (nullable = true)\n",
      " |-- plot_synopsis: string (nullable = true)\n",
      " |-- tags: string (nullable = true)\n",
      " |-- split: string (nullable = true)\n",
      " |-- synopsis_source: string (nullable = true)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# reading movie synopsis data\n",
    "\n",
    "df = spark.read.csv(\"gs://msca-bdp-student-gcs/movie_synopsis/mpst_full_data.csv\",\n",
    "                    header=True, inferSchema= True, sep=\",\", quote='\"', escape='\"', multiLine=True)\n",
    "\n",
    "df.printSchema()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "73971f7d-62bc-48ff-95be-29d4cbe95183",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                                \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---------+--------------------+--------------------+--------------------+-----+---------------+\n",
      "|  imdb_id|               title|       plot_synopsis|                tags|split|synopsis_source|\n",
      "+---------+--------------------+--------------------+--------------------+-----+---------------+\n",
      "|tt0057603|I tre volti della...|Note: this synops...|cult, horror, got...|train|           imdb|\n",
      "|tt1733125|Dungeons & Dragon...|Two thousand year...|            violence|train|           imdb|\n",
      "|tt0033045|The Shop Around t...|Matuschek's, a gi...|            romantic| test|           imdb|\n",
      "|tt0113862|  Mr. Holland's Opus|Glenn Holland, no...|inspiring, romant...|train|           imdb|\n",
      "|tt0086250|            Scarface|In May 1980, a Cu...|cruelty, murder, ...|  val|           imdb|\n",
      "+---------+--------------------+--------------------+--------------------+-----+---------------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df.show(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "871c1a8c-7390-4dc7-acee-81e3e3e76d1e",
   "metadata": {},
   "source": [
    "***b. Print basic summary statistics of the data set***\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "41bd5cb1-bff3-4f1b-851b-fad31197e3d2",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Stage 3:>                                                          (0 + 1) / 1]\r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------+---------+-----------------+-------------------------+--------------------+-----+---------------+\n",
      "|summary|  imdb_id|            title|            plot_synopsis|                tags|split|synopsis_source|\n",
      "+-------+---------+-----------------+-------------------------+--------------------+-----+---------------+\n",
      "|  count|    14828|            14828|                    14828|               14828|14828|          14828|\n",
      "|   mean|     null|615.7078947368421|                     null|                null| null|           null|\n",
      "| stddev|     null|820.9172438097585|                     null|                null| null|           null|\n",
      "|    min|tt0000091|                $|     \".hack//Roots\" fo...|              absurd| test|           imdb|\n",
      "|    max|tt6583664|        Üç Maymun|《龍虎門》Dragon Tiger...|whimsical, violen...|  val|      wikipedia|\n",
      "+-------+---------+-----------------+-------------------------+--------------------+-----+---------------+\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                                \r"
     ]
    }
   ],
   "source": [
    "df.describe().show() "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "03ac52a3-5f72-4f3d-b156-ba5dd8318be4",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                                \r"
     ]
    },
    {
     "data": {
      "text/plain": [
       "71"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.select(F.explode(F.split(F.col('tags'), ', '))).distinct().count()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5662af27-e527-41ac-9d35-cda60a989736",
   "metadata": {},
   "source": [
    "***c. Create a horizontal bar chart of movie tag counts such as action, cult, horror, etc.***"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "8ce4ee6a-6ef4-48d5-8648-9c496d2efd9b",
   "metadata": {},
   "outputs": [],
   "source": [
    "ind_tag = df.select(F.explode(F.split(F.col('tags'), ', ')).alias('tags'))\\\n",
    "            .groupby('tags').count()\\\n",
    "            .sort('count')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "1212f6f6-ea2a-4b02-8de3-6985c9302c68",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>tags</th>\n",
       "      <th>count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>non fiction</td>\n",
       "      <td>37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>christian film</td>\n",
       "      <td>42</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>autobiographical</td>\n",
       "      <td>44</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>magical realism</td>\n",
       "      <td>54</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>suicidal</td>\n",
       "      <td>54</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               tags  count\n",
       "0       non fiction     37\n",
       "1    christian film     42\n",
       "2  autobiographical     44\n",
       "3   magical realism     54\n",
       "4          suicidal     54"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tags_pd = ind_tag.toPandas()\n",
    "tags_pd.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "ef5c8b0d-76fd-46e8-95b2-59641bae982f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x2000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "tags_pd.plot(figsize=(10, 20),kind='barh', y='count', x='tags', grid=True, legend=False, color='#00796B')\n",
    "plt.xlabel('Count')\n",
    "plt.ylabel('Tag')\n",
    "plt.title('Movie Tag Counts')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c6be85c2-14d2-4b9f-9c3b-6ed151800cfb",
   "metadata": {},
   "source": [
    "***d. Arrange bar in descending order and identify the top 10 popular tags***"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "a284248a-2607-4cdd-9fd6-f5c5bf855528",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA7IAAAJuCAYAAACNCz4UAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAA9hAAAPYQGoP6dpAABrpklEQVR4nO3dd3gUZfv+/3PTNyEFQgsQemhKL1KkdxRBLKBRiCiodBFBREKQjiKIPHZpShGlfBCiEOkI0iNIkxaRoqEmmIQQsvP7wy/7c03AAEmWyb5fx8FBduaee67da5+HnN6zsxbDMAwBAAAAAGASbs4uAAAAAACA20GQBQAAAACYCkEWAAAAAGAqBFkAAAAAgKkQZAEAAAAApkKQBQAAAACYCkEWAAAAAGAqBFkAAAAAgKkQZAEAAAAApkKQBQAgiywWS5b+rF+/PsdqiIiIyFINEREROVbDv2vx9/fXX3/9lWH/b7/9Jjc3N1ksFkVFReVYHevXr8+21/1een0BADfn4ewCAAAwi61btzo8HjNmjNatW6e1a9c6bK9SpUqO1TBy5Ei99NJL9se7d+9W3759NX78eDVv3ty+vVChQjlWwz95enrq+vXr+uqrr/T888877Js1a5b8/f2VmJiYozXUqlVLW7duzZbX/V57fQEAmSPIAgCQRfXr13d4XKhQIbm5uWXYnpPKlSuncuXK2R9fvXpVkhQWFparddzg5eWljh07aubMmQ5B1jAMzZ49W127dtWnn36aozUEBARk23O/115fAEDmuLQYAIBsdPHiRfXp00fFixeXl5eXypYtqxEjRig1NdVhnMViUb9+/fTxxx+rQoUK8vb2VpUqVbRw4cJsqeOHH35Qy5YtFRAQIF9fXzVq1Ehr1qxxGBMVFSWLxaL9+/frqaeeUmBgoIoUKaKePXsqISEhy+fq2bOntmzZosOHDzuc/7ffftNzzz2X6TG//PKLOnXqpPz588vHx0c1atTQnDlz7PvPnTsnLy8vjRw5MsOxhw4dksVi0fTp0yXd/NLinTt36pFHHlGBAgXk4+OjmjVratGiRVl+Xjezc+dOdevWTaVLl5bValXp0qX11FNP6bfffsswdvPmzWrQoIF8fHxUvHhxjRw5Up999pksFovi4uLs49auXatmzZopODhYVqtVJUuW1GOPPabk5OS7rhcA8iKCLAAA2eTq1atq3ry55s6dq8GDB2vlypV65plnNHnyZHXp0iXD+OXLl2v69Ol666239M0336hUqVJ66qmn9M0339xVHV9++aXatGmjgIAAzZkzR4sWLVKBAgXUtm3bDGFWkh577DFVqFBBixcv1uuvv6758+frlVdeyfL5WrVqpVKlSmnmzJn2bZ9//rmaNGmisLCwDOMPHz6shg0bav/+/Zo+fbqWLFmiKlWqKCIiQpMnT5b092r3ww8/rDlz5shmszkcP2vWLHl5eSk8PPymNa1bt06NGjXS5cuX9dFHH+n//u//VKNGDXXt2lWzZ8/O8nPLTFxcnCpWrKhp06Zp1apVmjRpks6ePau6devq/Pnz9nF79+5V69atlZycrDlz5uijjz7S7t27NW7cuAzzPfTQQ/Ly8tLMmTP1/fffa+LEifLz89O1a9fuqlYAyLMMAABwR3r06GH4+fnZH3/00UeGJGPRokUO4yZNmmRIMlavXm3fJsmwWq3GH3/8Yd92/fp1o1KlSkb58uWzXMO6desMScbXX39tGIZhJCUlGQUKFDA6duzoMC49Pd2oXr26Ua9ePfu2UaNGGZKMyZMnO4zt06eP4ePjY9hstlue+5/Pf9SoUUbRokWNtLQ048KFC4a3t7cxe/Zs49y5c4YkY9SoUfbjunXrZnh7exsnT550mK99+/aGr6+vcfnyZcMwDGP58uUZXrfr168bxYoVMx577LEMr8G6devs2ypVqmTUrFnTSEtLczjHww8/bISEhBjp6em3fG7/nvvG65uZ69evG3/99Zfh5+dnvPfee/btTzzxhOHn52ecO3fOvi09Pd2oUqWKIck4ceKEYRiG8c033xiSjNjY2CzVBAAwDFZkAQDIJmvXrpWfn58ef/xxh+037nD779XQli1bqkiRIvbH7u7u6tq1q44ePapTp07dUQ1btmzRxYsX1aNHD12/ft3+x2azqV27dtqxY4eSkpIcjnnkkUccHlerVk1Xr15VfHx8ls/73HPP6c8//9R3332nefPmycvLS0888USmY9euXauWLVsqNDTUYXtERISSk5PtN9Vq3769ihYtqlmzZtnHrFq1SmfOnFHPnj1vWsvRo0d16NAh+4rtP1+HDh066OzZsw6XQd+uv/76S8OGDVP58uXl4eEhDw8P5cuXT0lJSTp48KB93IYNG9SiRQsVLFjQvs3NzU1PPvmkw3w1atSQl5eXevfurTlz5uj48eN3XBsAuApu9gQAQDa5cOGCihYtKovF4rC9cOHC8vDw0IULFxy2Fy1aNMMcN7ZduHBBJUqUuO0a/vzzT0nKEKb/6eLFi/Lz87M/Dg4Odtjv7e0tSUpJScnyeUuVKqWWLVtq5syZiouLU7du3eTr65vpZzwvXLigkJCQDNuLFStm3y9JHh4eevbZZ/X+++/r8uXLCgoK0uzZsxUSEqK2bdvetJYbr8GQIUM0ZMiQTMf88xLg2/X0009rzZo1GjlypOrWrauAgABZLBZ16NDB4TW7cOGCw3+ouOHf28qVK6cffvhBkydPVt++fZWUlKSyZctqwIABGjhw4B3XCQB5GUEWAIBsEhwcrG3btskwDIcwGx8fr+vXrzuszEnSH3/8kWGOG9v+HS6z6sY53n///ZveZTezcJUdevbsqWeeeUY2m00ffvjhTccFBwfr7NmzGbafOXNGkhxep+eee05vv/22Fi5cqK5du2r58uUaNGiQ3N3dbzr/jeOHDx+e6WeTJalixYpZek7/lpCQoBUrVmjUqFF6/fXX7dtTU1N18eJFh7HBwcH2UP1PmfW9cePGaty4sdLT07Vz5069//77GjRokIoUKaJu3brdUa0AkJcRZAEAyCYtW7bUokWLtGzZMj366KP27XPnzrXv/6c1a9bozz//tAfL9PR0ffXVVypXrtwdrcZKUqNGjRQUFKQDBw6oX79+d/hM7syjjz6qRx99VIGBgbf8qpqWLVtq6dKlOnPmjH0VVvr7dfL19XU4tnLlynrggQc0a9YspaenKzU19aZ3Qr6hYsWKCgsL088//6zx48ff/RP7B4vFIsMw7KvWN3z22WdKT0932Na0aVNFR0fr/Pnz9nBts9n09ddf33R+d3d3PfDAA6pUqZLmzZun3bt3E2QBIBMEWQAAskn37t31v//9Tz169FBcXJyqVq2qzZs3a/z48erQoYNatWrlML5gwYJq0aKFRo4cKT8/P33wwQc6dOjQXX0FT758+fT++++rR48eunjxoh5//HEVLlxY586d088//6xz587dcrX0bvj4+GTpjsujRo3SihUr1Lx5c0VGRqpAgQKaN2+eVq5cqcmTJyswMNBhfM+ePfXiiy/qzJkzatiwYZZWUz/++GO1b99ebdu2VUREhIoXL66LFy/q4MGD2r179y3D5K0EBASoSZMmevvtt1WwYEGVLl1aGzZs0Oeff66goCCHsSNGjNC3336rli1basSIEbJarfroo4/sn1F2c/v7ViUfffSR1q5dq4ceekglS5bU1atX7XeA/vd7BgDwN272BABANvHx8dG6desUHh6ut99+W+3bt9fs2bM1ZMgQLVmyJMP4Rx55RP369dObb76pxx57THFxcZo3b566du16V3U888wzWrdunf766y+9+OKLatWqlQYOHKjdu3dnWBV2hooVK2rLli2qWLGi+vbtq86dO+uXX37RrFmz9Nprr2UY361bN1mtVp06deo/V2NvaN68ubZv366goCANGjRIrVq10ssvv6wffvjhrsPh/Pnz1bx5cw0dOlRdunTRzp07FRMTkyGAV69eXTExMbJarerevbt69+6t++67T3369JEk+/gaNWro+vXrGjVqlNq3b69nn31W586d0/Lly9WmTZu7qhUA8iqLYRiGs4sAAMDVWCwW9e3bVzNmzHB2Kchlbdq0UVxcnH799VdnlwIApsWlxQAAADlk8ODBqlmzpkJDQ3Xx4kXNmzdPMTEx+vzzz51dGgCYGkEWAAAgh6SnpysyMlJ//PGHLBaLqlSpoi+++ELPPPOMs0sDAFPj0mIAAAAAgKlwsycAAAAAgKkQZAEAAAAApkKQBQAAAACYCjd7Qo6y2Ww6c+aM/P39ZbFYnF0OAAAAACcxDENXrlxRsWLF5OZ2d2uqBFnkqDNnzig0NNTZZQAAAAC4R/z+++8qUaLEXc1BkEWO8vf3lySdOHFCBQoUcHI1yE1paWlavXq12rRpI09PT2eXg1xG/10b/Xdd9N610X/XlpX+JyYmKjQ01J4R7gZBFjnqxuXE/v7+CggIcHI1yE1paWny9fVVQEAA/5i5IPrv2ui/66L3ro3+u7bb6X92fOSQmz0BAAAAAEyFIAsAAAAAMBWCLAAAAADAVAiyAAAAAABTIcgCAAAAAEyFIAsAAAAAMBWCLAAAAADAVAiyAAAAAABTIcgCAAAAAEyFIAsAAAAAMBUPZxcA11Bi0ghd9eDt5kqs7u5aUKeFAt8aopT0dGeXg1xG/10b/Xdd9N610f97jzF+hrNLyDGsyAIAAAAATIUgCwAAAAAwFYIsAAAAAMBUCLIAAAAAAFMhyAIAAAAATIUgCwAAAAAwFYIsAAAAAMBUCLIAAAAAAFMhyLogi8WiZcuWObsMAAAAALgjBFkAAAAAgKkQZPOotLS0HJv72rVrOTY3AAAAAPwXgqwTNGvWTP3799egQYOUP39+FSlSRJ988omSkpL03HPPyd/fX+XKldN3330nSZo9e7aCgoIc5li2bJksFov9cVRUlGrUqKGZM2eqbNmy8vb2lmEYOnLkiJo0aSIfHx9VqVJFMTExGeo5ffq0unbtqvz58ys4OFidOnVSXFycfX9ERIQ6d+6sCRMmqFixYqpQoUKOvC4AAAAAkBUezi7AVc2ZM0dDhw7V9u3b9dVXX+nll1/WsmXL9Oijj+qNN97Q1KlT9eyzz+rkyZNZnvPo0aNatGiRFi9eLHd3d9lsNnXp0kUFCxbUTz/9pMTERA0aNMjhmOTkZDVv3lyNGzfWxo0b5eHhobFjx6pdu3bau3evvLy8JElr1qxRQECAYmJiZBjGTWtITU1Vamqq/XFiYqIkyermLou7+228QjA7q5u7w99wLfTftdF/10XvXRv9v/fk5FWaNzvXrc6ZnfUQZJ2kevXqevPNNyVJw4cP18SJE1WwYEH16tVLkhQZGakPP/xQe/fuzfKc165d0xdffKFChQpJklavXq2DBw8qLi5OJUqUkCSNHz9e7du3tx+zcOFCubm56bPPPrOv8M6aNUtBQUFav3692rRpI0ny8/PTZ599Zg+2NzNhwgSNHj06w/b/VX9Qvr6+WX4uyDtm1mrq7BLgRPTftdF/10XvXRv9v3dER0fn+jkzuwL0huTk5Gw7D0HWSapVq2b/2d3dXcHBwapatap9W5EiRSRJ8fHxWZ6zVKlS9hArSQcPHlTJkiXtIVaSGjRo4HDMrl27dPToUfn7+ztsv3r1qo4dO2Z/XLVq1f8MsdLfoXzw4MH2x4mJiQoNDVXfnzfrqidvN1didXPXzFpN1XP3BqXY0p1dDnIZ/Xdt9N910XvXRv/vPQmR7+TaudLS0hQTE6PWrVvL09Mz0zE3rtbMDiQLJ/l3cy0Wi8O2G6ujNptNbm5uGS7nzWxZ3s/Pz+FxZpcA//NztTfmr127tubNm5dh7D9D8b/nvhlvb295e3tn2J5iS9fVdEsmRyCvS7GlKyWdf8xcFf13bfTfddF710b/7x03C5Q5fc6bnTc76yHImkChQoV05coVJSUl2QNlbGzsfx5XpUoVnTx5UmfOnFGxYsUkSVu3bnUYU6tWLX311VcqXLiwAgICsr12AAAAAMhu3LXYBB544AH5+vrqjTfe0NGjRzV//nzNnj37P49r1aqVKlasqO7du+vnn3/Wpk2bNGLECIcx4eHhKliwoDp16qRNmzbpxIkT2rBhgwYOHKhTp07l0DMCAAAAgDtHkDWBAgUK6Msvv1R0dLSqVq2qBQsWKCoq6j+Pc3Nz09KlS5Wamqp69erphRde0Lhx4xzG+Pr6auPGjSpZsqS6dOmiypUrq2fPnkpJSWGFFgAAAMA9iUuLnWD9+vUZtv3ze1tv+OdnXDt37qzOnTs77L9xh2Pp7++RzSzcVqhQQZs2bbrpvJJUtGhRzZkz56b1ZmX1FwAAAAByCyuyAAAAAABTIcgCAAAAAEyFIAsAAAAAMBWCLAAAAADAVAiyAAAAAABTIcgCAAAAAEyFr99Brjg1bJyCg4OdXQZyUVpamqKjo5UQ+Y48PT2dXQ5yGf13bfTfddF710b/kZtYkQUAAAAAmApBFgAAAABgKgRZAAAAAICpEGQBAAAAAKZCkAUAAAAAmApBFgAAAABgKgRZAAAAAICpEGQBAAAAAKZCkAUAAAAAmApBFgAAAABgKgRZAAAAAICpEGQBAAAAAKZCkAUAAAAAmApBFgAAAABgKgRZAAAAAICpEGQBAAAAAKZCkAUAAAAAmApBFgAAAABgKgRZAAAAAICpEGQBAAAAAKZCkAUAAAAAmApBFgAAAABgKgRZAAAAAICpEGQBAAAAAKZCkAUAAAAAmApBFgAAAABgKh7OLgCuocSkEbrqwdvNlVjd3bWgTgsFvjVEKenpzi4HuYz+uzb677ryeu+N8TOcXQKA/4cVWQAAAACAqRBkAQAAAACmQpAFAAAAAJgKQRYAAAAAYCoEWQAAAACAqRBkAQAAAACmQpAFAAAAAJgKQRYAAAAAYCoEWSeJiopSjRo1sjw+Li5OFotFsbGxOVYTAAAAAJgBQdZJhgwZojVr1ji7DAAAAAAwHQ9nF+Cq8uXLp3z58jm7DAAAAAAwHVZkc8jHH3+s4sWLy2azOWx/5JFH1KNHjwyXFttsNr311lsqUaKEvL29VaNGDX3//fe3PMeBAwfUoUMH5cuXT0WKFNGzzz6r8+fP2/c3a9ZMAwYM0NChQ1WgQAEVLVpUUVFRDnNcvnxZvXv3VpEiReTj46P7779fK1assO/fsmWLmjRpIqvVqtDQUA0YMEBJSUl3/sIAAAAAwF1iRTaHPPHEExowYIDWrVunli1bSpIuXbqkVatW6dtvv9WWLVscxr/33nuaMmWKPv74Y9WsWVMzZ87UI488ov379yssLCzD/GfPnlXTpk3Vq1cvvfvuu0pJSdGwYcP05JNPau3atfZxc+bM0eDBg7Vt2zZt3bpVERERatSokVq3bi2bzab27dvrypUr+vLLL1WuXDkdOHBA7u7ukqR9+/apbdu2GjNmjD7//HOdO3dO/fr1U79+/TRr1qxMn3dqaqpSU1PtjxMTEyVJVjd3Wf7fvHANVjd3h7/hWui/a6P/riuv9z4tLc3ZJdzTbrw+vE6uKSv9z873hsUwDCPbZoODTp06qWDBgvr8888lSZ988olGjRqlU6dOacyYMVq2bJn95k3FixdX37599cYbb9iPr1evnurWrav//e9/iouLU5kyZbRnzx7VqFFDkZGR2rZtm1atWmUff+rUKYWGhurw4cOqUKGCmjVrpvT0dG3atMlhzhYtWmjixIlavXq12rdvr4MHD6pChQoZ6u/evbusVqs+/vhj+7bNmzeradOmSkpKko+PT4ZjoqKiNHr06Azb58+fL19f39t/EQEAAADkCcnJyXr66aeVkJCggICAu5qLFdkcFB4ert69e+uDDz6Qt7e35s2bp27dutlXPG9ITEzUmTNn1KhRI4ftjRo10s8//5zp3Lt27dK6desy/ZztsWPH7MG0WrVqDvtCQkIUHx8vSYqNjVWJEiUyDbE3znH06FHNmzfPvs0wDNlsNp04cUKVK1fOcMzw4cM1ePBgh+cWGhqqvj9v1lVP3m6uxOrmrpm1mqrn7g1KsaU7uxzkMvrv2ui/68rrvU+IfMfZJdzT0tLSFBMTo9atW8vT09PZ5SCXZaX/N67WzA4kixzUsWNH2Ww2rVy5UnXr1tWmTZv07rvv3nS8xWJxeGwYRoZtN9hsNnXs2FGTJk3KsC8kJMT+87/fRBaLxf65XavVesv6bTabXnzxRQ0YMCDDvpIlS2Z6jLe3t7y9vTNsT7Gl62p65s8FeVuKLV0p6XnvlxlkDf13bfTfdeXV3hPOssbT05PXyoXdqv/Z+b4gyOYgq9WqLl26aN68eTp69KgqVKig2rVrZxgXEBCgYsWKafPmzWrSpIl9+5YtW1SvXr1M565Vq5YWL16s0qVLy8PjztpYrVo1nTp1Sr/++mumq7K1atXS/v37Vb58+TuaHwAAAAByAnctzmHh4eFauXKlZs6cqWeeeeam41577TVNmjRJX331lQ4fPqzXX39dsbGxGjhwYKbj+/btq4sXL+qpp57S9u3bdfz4ca1evVo9e/ZUehb/C2jTpk3VpEkTPfbYY4qJidGJEyf03Xff2e+WPGzYMG3dulV9+/ZVbGysjhw5ouXLl6t///63/0IAAAAAQDZhRTaHtWjRQgUKFNDhw4f19NNP33TcgAEDlJiYqFdffVXx8fGqUqWKli9fnukdiyWpWLFi+vHHHzVs2DC1bdtWqampKlWqlNq1ayc3t6z/94nFixdryJAheuqpp5SUlKTy5ctr4sSJkv5esd2wYYNGjBihxo0byzAMlStXTl27dr29FwEAAAAAshFBNoe5u7vrzJkzGbZHRUU5fKerm5ubIiMjFRkZmek8pUuX1r9vMB0WFqYlS5bc9Nzr16/PsG3ZsmUOjwsUKKCZM2fedI66detq9erVN90PAAAAALmNS4sBAAAAAKZCkAUAAAAAmApBFgAAAABgKgRZAAAAAICpEGQBAAAAAKZCkAUAAAAAmApfv4NccWrYOAUHBzu7DOSitLQ0RUdHKyHyHXl6ejq7HOQy+u/a6L/rovcAcgsrsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMxcPZBcA1lJg0Qlc9eLu5Equ7uxbUaaHAt4YoJT3d2eUgl9H/vM8YP8PZJQAAXBgrsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsnfIMAz17t1bBQoUkMViUVBQkAYNGpQtc8fFxclisSg2NvamY2bPnq2goKBsOd/NrF+/XhaLRZcvX87R8wAAAADA7SDI3qHvv/9es2fP1ooVK3T27Fndf//9zi4JAAAAAFyCh7MLMKtjx44pJCREDRs2lCR5ePBSAgAAAEBuIH3dgYiICM2ZM0eSZLFYVKpUKZUuXdphzJdffqlp06bp8OHD8vPzU4sWLTRt2jQVLlxYknTp0iX169dPq1ev1l9//aUSJUrojTfe0HPPPWef4/jx43rllVe0bds2hYWF6aOPPlKDBg0czrNs2TINHTpUJ0+eVOPGjTVz5kyFhoZK+jtsDx48WD/99JOSkpJUuXJlTZgwQa1atbIfn5qaqpEjR2rBggWKj49XyZIl9frrr+v555/P8LxTUlL0+OOP68KFC4qOjlaBAgUyjElNTVVqaqr9cWJioiTJ6uYui7v77bzMMDmrm7vD33At9D/vS0tL+899txqDvIneuzb679qy0v/sfG8QZO/Ae++9p3LlyumTTz7Rjh075O7urieeeMJhzLVr1zRmzBhVrFhR8fHxeuWVVxQREaHo6GhJ0siRI3XgwAF99913KliwoI4ePaqUlBSHOUaMGKF33nlHYWFhGjFihJ566ikdPXrUvvqbnJyscePGac6cOfLy8lKfPn3UrVs3/fjjj5Kkv/76Sx06dNDYsWPl4+OjOXPmqGPHjjp8+LBKliwpSerevbu2bt2q6dOnq3r16jpx4oTOnz+f4TknJCTo4Ycflo+Pj9asWSM/P79MX5sJEyZo9OjRGbb/r/qD8vX1vc1XGnnBzFpNnV0CnIj+5103/j27lZiYmFyoBPcieu/a6L9ru1X/k5OTs+08BNk7EBgYKH9/f7m7u6to0aKZjunZs6f957Jly2r69OmqV6+e/vrrL+XLl08nT55UzZo1VadOHUnKsKIrSUOGDNFDDz0kSRo9erTuu+8+HT16VJUqVZL093/RmDFjhh544AFJ0pw5c1S5cmVt375d9erVU/Xq1VW9enX7fGPHjtXSpUu1fPly9evXT7/++qsWLVqkmJgY+ypt2bJlM9Tx559/qmvXripXrpwWLFggLy+vm742w4cP1+DBg+2PExMTFRoaqr4/b9ZVT95ursTq5q6ZtZqq5+4NSrGlO7sc5DL6n/clRL5z031paWmKiYlR69at5enpmYtVwdnovWuj/64tK/2/cbVmdiBZ5JA9e/YoKipKsbGxunjxomw2myTp5MmTqlKlil5++WU99thj2r17t9q0aaPOnTvbP297Q7Vq1ew/h4SESJLi4+PtQdbDw8MehCWpUqVKCgoK0sGDB1WvXj0lJSVp9OjRWrFihc6cOaPr168rJSVFJ0+elCTFxsbK3d1dTZveesWkVatWqlu3rhYtWiT3/7g82NvbW97e3hm2p9jSdTXdcstjkTel2NKVkk6QcVX0P+/Kyi+pnp6e/DLroui9a6P/ru1W/c/O9wV3Lc4BSUlJatOmjfLly6cvv/xSO3bs0NKlSyX9fcmxJLVv316//fabBg0apDNnzqhly5YaMmSIwzz/bLTF8ncIvBGI/709s22vvfaaFi9erHHjxmnTpk2KjY1V1apV7TVYrdYsPZ+HHnpImzZt0oEDB7I0HgAAAAByEkE2Bxw6dEjnz5/XxIkT1bhxY1WqVEnx8fEZxhUqVEgRERH2G0N98sknt3We69eva+fOnfbHhw8f1uXLl+0rtps2bVJERIQeffRRVa1aVUWLFlVcXJx9fNWqVWWz2bRhw4ZbnmfixInq0aOHWrZsSZgFAAAA4HQE2RxQsmRJeXl56f3339fx48e1fPlyjRkzxmFMZGSk/u///k9Hjx7V/v37tWLFClWuXPm2zuPp6an+/ftr27Zt2r17t5577jnVr19f9erVkySVL19eS5YsUWxsrH7++Wc9/fTTDiu6pUuXVo8ePdSzZ08tW7ZMJ06c0Pr167Vo0aIM53rnnXcUHh6uFi1a6NChQ3fwqgAAAABA9iDI5oBChQpp9uzZ+vrrr1WlShVNnDhR77zjeFMMLy8vDR8+XNWqVVOTJk3k7u6uhQsX3tZ5fH19NWzYMD399NNq0KCBrFarwxxTp05V/vz51bBhQ3Xs2FFt27ZVrVq1HOb48MMP9fjjj6tPnz6qVKmSevXqpaSkpEzPN3XqVD355JNq0aKFfv3119uqFQAAAACyi8UwDMPZRSDvSkxMVGBgoHxee1FXPbi3mCuxurtrQZ0WemrnWm7244Lof95njJ9x031paWmKjo5Whw4duOGLi6H3ro3+u7as9P9GNkhISFBAQMBdnY8VWQAAAACAqRBkAQAAAACmQpAFAAAAAJgKQRYAAAAAYCoEWQAAAACAqRBkAQAAAACmwvehIFecGjZOwcHBzi4DuejGLdgTIt/hFvwuiP4DAICcxIosAAAAAMBUCLIAAAAAAFMhyAIAAAAATIUgCwAAAAAwFYIsAAAAAMBUCLIAAAAAAFMhyAIAAAAATIUgCwAAAAAwFYIsAAAAAMBUCLIAAAAAAFMhyAIAAAAATIUgCwAAAAAwFYIsAAAAAMBUCLIAAAAAAFMhyAIAAAAATIUgCwAAAAAwFYIsAAAAAMBUCLIAAAAAAFMhyAIAAAAATIUgCwAAAAAwFYIsAAAAAMBUCLIAAAAAAFMhyAIAAAAATIUgCwAAAAAwFYIsAAAAAMBUCLIAAAAAAFPxcHYBcA0lJo3QVQ/ebq7E6u6uBXVaKPCtIUpJT3d2Ochl9D/vMcbPcHYJAADYsSILAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIwm727NkKCgpydhkAAAAAcEsEWRdVunRpTZs2zWFb165d9euvvzqnIAAAAADIIg9nF5BXXbt2TV5eXs4u47ZYrVZZrVZnlwEAAAAAt8SKbDZp1qyZ+vXrp8GDB6tgwYJq3bq1NmzYoHr16snb21shISF6/fXXdf36dYdj+vfvr0GDBil//vwqUqSIPvnkEyUlJem5556Tv7+/ypUrp++++85+THp6up5//nmVKVNGVqtVFStW1HvvvedQS0REhDp37qx33nlHISEhCg4OVt++fZWWlmY/72+//aZXXnlFFotFFotFUuaXFi9fvlx16tSRj4+PChYsqC5duuTQKwgAAAAAWcOKbDaaM2eOXn75Zf344486f/682rRpo4iICM2dO1eHDh1Sr1695OPjo6ioKIdjhg4dqu3bt+urr77Syy+/rGXLlunRRx/VG2+8oalTp+rZZ5/VyZMn5evrK5vNphIlSmjRokUqWLCgtmzZot69eyskJERPPvmkfd5169YpJCRE69at09GjR9W1a1fVqFFDvXr10pIlS1S9enX17t1bvXr1uunzWblypbp06aIRI0boiy++0LVr17Ry5cpbvgapqalKTU21P05MTJQkWd3cZXF3v8NXFmZkdXN3+Buuhf7nPTf+Y+jtjL2dY5A30HvXRv9dW1b6n53vDYthGEa2zebCmjVrpoSEBO3Zs0eSNGLECC1evFgHDx60r3h+8MEHGjZsmBISEuTm5qZmzZopPT1dmzZtkvT3amtgYKC6dOmiuXPnSpL++OMPhYSEaOvWrapfv36m5+7bt6/+/PNPffPNN5L+XpFdv369jh07Jvf/Fx6ffPJJubm5aeHChZL+/ozsoEGDNGjQIPs8s2fP1qBBg3T58mVJUsOGDVW2bFl9+eWXWX4doqKiNHr06Azb58+fL19f3yzPAwAAACBvSU5O1tNPP62EhAQFBATc1VysyGajOnXq2H8+ePCgGjRoYA+xktSoUSP99ddfOnXqlEqWLClJqlatmn2/u7u7goODVbVqVfu2IkWKSJLi4+Pt2z766CN99tln+u2335SSkqJr166pRo0aDrXcd9999hArSSEhIdq3b99tPZ/Y2NhbrthmZvjw4Ro8eLD9cWJiokJDQ9X358266snbzZVY3dw1s1ZT9dy9QSm2dGeXg1xG//OehMh3sjw2LS1NMTExat26tTw9PXOwKtxr6L1ro/+uLSv9v3G1ZnYgWWQjPz8/+8+GYTiE2BvbJDls/3eTLRaLw7YbY202myRp0aJFeuWVVzRlyhQ1aNBA/v7+evvtt7Vt2zaHeTKb98YcWXUnN37y9vaWt7d3hu0ptnRdTbdkcgTyuhRbulLSCTKuiv7nHXfyS6mnpye/zLooeu/a6L9ru1X/s/N9wc2eckiVKlW0ZcsW/fPK7S1btsjf31/Fixe/43k3bdqkhg0bqk+fPqpZs6bKly+vY8eO3fY8Xl5eSv+PXy6rVaumNWvW3GmpAAAAAJAjCLI5pE+fPvr999/Vv39/HTp0SP/3f/+nUaNGafDgwXJzu/OXvXz58tq5c6dWrVqlX3/9VSNHjtSOHTtue57SpUtr48aNOn36tM6fP5/pmFGjRmnBggUaNWqUDh48qH379mny5Ml3XDsAAAAAZAeCbA4pXry4oqOjtX37dlWvXl0vvfSSnn/+eb355pt3Ne9LL72kLl26qGvXrnrggQd04cIF9enT57bneeuttxQXF6dy5cqpUKFCmY5p1qyZvv76ay1fvlw1atRQixYtMlzCDAAAAAC5jbsWI0clJiYqMDBQPq+9qKsefCTblVjd3bWgTgs9tXMtn5F0QfQ/7zHGz8jy2LS0NEVHR6tDhw58Ts7F0HvXRv9dW1b6fyMbZMddi1mRBQAAAACYCkEWAAAAAGAqBFkAAAAAgKkQZAEAAAAApkKQBQAAAACYCkEWAAAAAGAqBFkAAAAAgKnwxZ7IFaeGjVNwcLCzy0AuuvFdYgmR7/Bdci6I/gMAgJzEiiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAU/FwdgFwDSUmjdBVD95ursTq7q4FdVoo8K0hSklPd3Y5yGX0P3sZ42c4uwQAAO4prMgCAAAAAEyFIAsAAAAAMBWCLAAAAADAVAiyAAAAAABTIcgCAAAAAEyFIAsAAAAAMBWCLAAAAADAVAiyyNTs2bMVFBTk7DIAAAAAIAOCLLIkKipKNWrUcHYZAAAAAECQBQAAAACYC0E2D7PZbJo0aZLKly8vb29vlSxZUuPGjdP69etlsVh0+fJl+9jY2FhZLBbFxcVlmGf27NkaPXq0fv75Z1ksFlksFs2ePTvXngcAAAAA/JOHswtAzhk+fLg+/fRTTZ06VQ8++KDOnj2rQ4cO3fY8Xbt21S+//KLvv/9eP/zwgyQpMDAw07GpqalKTU21P05MTJQkWd3cZXF3v4NnAbOyurk7/A3XQv+zV1pamrNLuC036jVb3bh79N610X/XlpX+Z+d7gyCbR125ckXvvfeeZsyYoR49ekiSypUrpwcffFDr16+/rbmsVqvy5csnDw8PFS1a9JZjJ0yYoNGjR2fY/r/qD8rX1/e2zou8YWatps4uAU5E/7NHdHS0s0u4IzExMc4uAU5C710b/Xdtt+p/cnJytp2HIJtHHTx4UKmpqWrZsmWunnf48OEaPHiw/XFiYqJCQ0PV9+fNuurJ282VWN3cNbNWU/XcvUEptnRnl4NcRv+zV0LkO84u4bakpaUpJiZGrVu3lqenp7PLQS6i966N/ru2rPT/xtWa2YFkkUdZrdab7nNz+/uj0YZh2Ldl1zK/t7e3vL29M2xPsaXrarolW84Bc0mxpSslnSDjquh/9jDrL4Senp6mrR13h967Nvrv2m7V/+x8X3CzpzwqLCxMVqtVa9asybCvUKFCkqSzZ8/at8XGxt5yPi8vL6XzyygAAACAewArsnmUj4+Phg0bpqFDh8rLy0uNGjXSuXPntH//fnXv3l2hoaGKiorS2LFjdeTIEU2ZMuWW85UuXVonTpxQbGysSpQoIX9//0xXXgEAAAAgp7Eim4eNHDlSr776qiIjI1W5cmV17dpV8fHx8vT01IIFC3To0CFVr15dkyZN0tixY28512OPPaZ27dqpefPmKlSokBYsWJBLzwIAAAAAHLEim4e5ublpxIgRGjFiRIZ9jRo10t69ex22/fMzsxEREYqIiLA/9vb21jfffJNjtQIAAABAVrEiCwAAAAAwFYIsAAAAAMBUCLIAAAAAAFMhyAIAAAAATIUgCwAAAAAwFYIsAAAAAMBUCLIAAAAAAFPhe2SRK04NG6fg4GBnl4FclJaWpujoaCVEviNPT09nl4NcRv8BAEBOYkUWAAAAAGAqBFkAAAAAgKkQZAEAAAAApnLbn5EdPHhwptstFot8fHxUvnx5derUSQUKFLjr4gAAAAAA+LfbDrJ79uzR7t27lZ6erooVK8owDB05ckTu7u6qVKmSPvjgA7366qvavHmzqlSpkhM1AwAAAABc2G1fWtypUye1atVKZ86c0a5du7R7926dPn1arVu31lNPPaXTp0+rSZMmeuWVV3KiXgAAAACAi7vtIPv2229rzJgxCggIsG8LCAhQVFSUJk+eLF9fX0VGRmrXrl3ZWigAAAAAANIdBNmEhATFx8dn2H7u3DklJiZKkoKCgnTt2rW7rw4AAAAAgH+5o0uLe/bsqaVLl+rUqVM6ffq0li5dqueff16dO3eWJG3fvl0VKlTI7loBAAAAALj9mz19/PHHeuWVV9StWzddv37970k8PNSjRw9NnTpVklSpUiV99tln2VspAAAAAAC6gyCbL18+ffrpp5o6daqOHz8uwzBUrlw55cuXzz6mRo0a2VkjAAAAAAB2tx1kb8iXL5+qVauWnbUAAAAAAPCf7ijI7tixQ19//bVOnjyZ4aZOS5YsyZbCAAAAAADIzG3f7GnhwoVq1KiRDhw4oKVLlyotLU0HDhzQ2rVrFRgYmBM1AgAAAABgd9tBdvz48Zo6dapWrFghLy8vvffeezp48KCefPJJlSxZMidqBAAAAADA7raD7LFjx/TQQw9Jkry9vZWUlCSLxaJXXnlFn3zySbYXCAAAAADAP912kC1QoICuXLkiSSpevLh++eUXSdLly5eVnJycvdUBAAAAAPAvWQ6yPXv21JUrV9S4cWPFxMRIkp588kkNHDhQvXr10lNPPaWWLVvmWKEAAAAAAEi3cdfiOXPmaOLEiZoxY4auXr0qSRo+fLg8PT21efNmdenSRSNHjsyxQgEAAAAAkG4jyBqGIenvS4tvcHNz09ChQzV06NDsrwwAAAAAgEzc1mdkLRZLTtUBAAAAAECWZHlFVpIqVKjwn2H24sWLd1UQAAAAAAC3cltBdvTo0QoMDMypWgAAAAAA+E+3FWS7deumwoUL51QtAAAAAAD8pyx/RpbPxwIAAAAA7gVZDrI37loMAAAAAIAzZfnSYpvNlpN1II8rMWmErnrc1pXsMDmru7sW1GmhwLeGKCU93dnlIJe5Yv+N8TOcXQIAAC7jtr5+BwAAAAAAZyPIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHI5qJr1645uwQAAAAAMD2CbA5q1qyZ+vXrp8GDB6tgwYJq3bq1Dhw4oA4dOihfvnwqUqSInn32WZ0/f16S9PHHH6t48eKy2WwO8zzyyCPq0aOH/fG3336r2rVry8fHR2XLltXo0aN1/fp1+36LxaLPPvtMjz76qHx9fRUWFqbly5c7zLl8+XKFhYXJarWqefPmmjNnjiwWiy5fvmwfs2XLFjVp0kRWq1WhoaEaMGCAkpKScuCVAgAAAICs83B2AXndnDlz9PLLL+vHH3/UxYsX1bRpU/Xq1UvvvvuuUlJSNGzYMD355JNau3atnnjiCQ0YMEDr1q1Ty5YtJUmXLl3SqlWr9O2330qSVq1apWeeeUbTp09X48aNdezYMfXu3VuSNGrUKPt5R48ercmTJ+vtt9/W+++/r/DwcP32228qUKCA4uLi9Pjjj2vgwIF64YUXtGfPHg0ZMsSh7n379qlt27YaM2aMPv/8c507d079+vVTv379NGvWrJs+39TUVKWmptofJyYmSpKsbu6yuLtnz4sKU7C6uTv8Ddfiiv1PS0tzdgn3jBuvBa+J66H3ro3+u7as9D873xsWwzCMbJsNDpo1a6aEhATt2bNHkhQZGalt27Zp1apV9jGnTp1SaGioDh8+rAoVKqhTp04qWLCgPv/8c0nSJ598olGjRunUqVNyd3dXkyZN1L59ew0fPtw+x5dffqmhQ4fqzJkzkv5ekX3zzTc1ZswYSVJSUpL8/f0VHR2tdu3a6fXXX9fKlSu1b98++xxvvvmmxo0bp0uXLikoKEjdu3eX1WrVxx9/bB+zefNmNW3aVElJSfLx8cn0OUdFRWn06NEZts+fP1++vr53+lICAAAAMLnk5GQ9/fTTSkhIUEBAwF3NxYpsDqtTp4795127dmndunXKly9fhnHHjh1ThQoVFB4ert69e+uDDz6Qt7e35s2bp27dusn9/61m7tq1Szt27NC4cePsx6anp+vq1atKTk62h8Vq1arZ9/v5+cnf31/x8fGSpMOHD6tu3boO569Xr57D4127duno0aOaN2+efZthGLLZbDpx4oQqV66c6fMdPny4Bg8ebH+cmJio0NBQ9f15s6568nZzJVY3d82s1VQ9d29Qii3d2eUgl7li/xMi33F2CfeMtLQ0xcTEqHXr1vL09HR2OchF9N610X/XlpX+37haMzuQLHKYn5+f/WebzaaOHTtq0qRJGcaFhIRIkjp27CibzaaVK1eqbt262rRpk959912HOUaPHq0uXbpkmOOfq6T/fvNYLBb7Z28Nw5DFYnHY/++FeZvNphdffFEDBgzIcJ6SJUve9Pl6e3vL29s7w/YUW7quplsyOQJ5XYotXSnprhFkkJEr9Z9f2jLy9PTkdXFR9N610X/Xdqv+Z+f7giCbi2rVqqXFixerdOnS8vDI/KW3Wq3q0qWL5s2bp6NHj6pChQqqXbu2wxyHDx9W+fLl77iOSpUqKTo62mHbzp07M9S6f//+uzoPAAAAAOQE7lqci/r27auLFy/qqaee0vbt23X8+HGtXr1aPXv2VPo/VizCw8O1cuVKzZw5U88884zDHJGRkZo7d66ioqK0f/9+HTx4UF999ZXefPPNLNfx4osv6tChQxo2bJh+/fVXLVq0SLNnz5Yk+0rtsGHDtHXrVvXt21exsbE6cuSIli9frv79+9/9CwEAAAAAd4Egm4uKFSumH3/8Uenp6Wrbtq3uv/9+DRw4UIGBgXJz+/9b0aJFCxUoUECHDx/W008/7TBH27ZttWLFCsXExKhu3bqqX7++3n33XZUqVSrLdZQpU0bffPONlixZomrVqunDDz/UiBEjJMl+WXC1atW0YcMGHTlyRI0bN1bNmjU1cuRI+yXQAAAAAOAsXFqcg9avX59hW1hYmJYsWXLL49zd3e13IM5M27Zt1bZt25vuz+xG1P/8fljp7++mfeSRR+yPx40bpxIlSjh8zrZu3bpavXr1LWsFAAAAgNxGkHVRH3zwgerWravg4GD9+OOPevvtt9WvXz9nlwUAAAAA/4kg66KOHDmisWPH6uLFiypZsqReffVVh++mBQAAAIB7FUHWRU2dOlVTp051dhkAAAAAcNu42RMAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVbvaEXHFq2DgFBwc7uwzkorS0NEVHRysh8h15eno6uxzkMvoPAAByEiuyAAAAAABTIcgCAAAAAEyFIAsAAAAAMBWCLAAAAADAVAiyAAAAAABTIcgCAAAAAEyFIAsAAAAAMBWCLAAAAADAVAiyAAAAAABTIcgCAAAAAEyFIAsAAAAAMBWCLAAAAADAVAiyAAAAAABTIcgCAAAAAEyFIAsAAAAAMBWCLAAAAADAVAiyAAAAAABTIcgCAAAAAEyFIAsAAAAAMBWCLAAAAADAVAiyAAAAAABTIcgCAAAAAEyFIAsAAAAAMBWCLAAAAADAVAiyAAAAAABTIcgCAAAAAEzFw9kFwDWUmDRCVz14u7kSq7u7FtRpocC3higlPd3Z5eAuGeNnOLsEAAAAO1ZkAQAAAACmQpAFAAAAAJgKQRYAAAAAYCoEWQAAAACAqRBkAQAAAACmQpAFAAAAAJgKQRYAAAAAYCoEWScoXbq0pk2blu3zRkREqHPnzlkev379elksFl2+fFmSNHv2bAUFBWV7XQAAAACQnQiysOvatat+/fVXZ5cBAAAAALfk4ewCcO+wWq2yWq3OLgMAAAAAbsllV2SbNWumfv36qV+/fgoKClJwcLDefPNNGYYhSfrggw8UFhYmHx8fFSlSRI8//rgkae7cuQoODlZqaqrDfI899pi6d+9uf7x8+XLVqVNHPj4+KliwoLp06eIwPjk5WT179pS/v79KliypTz75xGH/6dOn1bVrV+XPn1/BwcHq1KmT4uLi7PvT09M1ePBge+1Dhw61136DYRiaPHmyypYtK6vVqurVq+ubb7656WuS2aXF//U8AAAAACC3ufSK7Jw5c/T8889r27Zt2rlzp3r37q1SpUqpZs2aGjBggL744gs1bNhQFy9e1KZNmyRJTzzxhAYMGKDly5friSeekCSdP39eK1as0Pfffy9JWrlypbp06aIRI0boiy++0LVr17Ry5UqHc0+ZMkVjxozRG2+8oW+++UYvv/yymjRpokqVKik5OVnNmzdX48aNtXHjRnl4eGjs2LFq166d9u7dKy8vL02ZMkUzZ87U559/ripVqmjKlClaunSpWrRoYT/Hm2++qSVLlujDDz9UWFiYNm7cqGeeeUaFChVS06ZN//P1ycrz+LfU1FSHkJ+YmChJsrq5y+LunoWuIK+wurk7/A1zS0tLu6Pxt3sc8gb677rovWuj/64tK/3PzveGxfj3Mp6LaNasmeLj47V//35ZLBZJ0uuvv67ly5dr7Nixeu6553Tq1Cn5+/tnOLZPnz6Ki4tTdHS0JOm9997T9OnTdfToUVksFjVs2FBly5bVl19+mem5S5curcaNG+uLL76Q9PfKadGiRTV69Gi99NJLmjlzpiZPnqyDBw/aa7t27ZqCgoK0bNkytWnTRsWKFdPAgQM1bNgwSdL169dVpkwZ1a5dW8uWLVNSUpIKFiyotWvXqkGDBvZzv/DCC0pOTtb8+fO1fv16NW/eXJcuXVJQUJBmz56tQYMG2W/+9F/PIzNRUVEaPXp0hu3z58+Xr69vlucBAAAAkLckJyfr6aefVkJCggICAu5qLpdeka1fv749KEpSgwYNNGXKFLVs2VKlSpVS2bJl1a5dO7Vr106PPvqoPYj16tVLdevW1enTp1W8eHHNmjVLERER9rliY2PVq1evW567WrVq9p8tFouKFi2q+Ph4SdKuXbt09OjRDCH66tWrOnbsmBISEnT27FmHgOrh4aE6derYLy8+cOCArl69qtatWzvMce3aNdWsWTNLr09Wnse/DR8+XIMHD7Y/TkxMVGhoqPr+vFlXPV367eZyrG7umlmrqXru3qAUW7qzy8FdSoh857bGp6WlKSYmRq1bt5anp2cOVYV7Ff13XfTetdF/15aV/t+4WjM7kCwykS9fPu3evVvr16/X6tWrFRkZqaioKO3YsUNBQUGqWbOmqlevrrlz56pt27bat2+fvv32W/vxWblh0r+ba7FYZLPZJEk2m021a9fWvHnzMhxXqFChLD2HG3OtXLlSxYsXd9jn7e2dpTnu5MZP3t7emc6fYkvX1XRLJkcgr0uxpSslnSBrdnf6C4mnpye/zLgw+u+66L1ro/+u7Vb9z873hcve7EmSfvrppwyPw8LC5O7uLg8PD7Vq1UqTJ0/W3r17FRcXp7Vr19rHvvDCC5o1a5ZmzpypVq1aKTQ01L6vWrVqWrNmzR3XVatWLR05ckSFCxdW+fLlHf4EBgYqMDBQISEhDvVfv35du3btsj+uUqWKvL29dfLkyQxz/LPWW7nb5wEAAAAAOcGlV2R///13DR48WC+++KJ2796t999/X1OmTNGKFSt0/PhxNWnSRPnz51d0dLRsNpsqVqxoPzY8PFxDhgzRp59+qrlz5zrMO2rUKLVs2VLlypVTt27ddP36dX333XcaOnRoluoKDw/X22+/rU6dOumtt95SiRIldPLkSS1ZskSvvfaaSpQooYEDB2rixIkKCwtT5cqV9e6779o/2ypJ/v7+GjJkiF555RXZbDY9+OCDSkxM1JYtW5QvXz716NHjP+u42+cBAAAAADnBpVdku3fvrpSUFNWrV099+/ZV//791bt3bwUFBWnJkiVq0aKFKleurI8++kgLFizQfffdZz82ICBAjz32mPLly6fOnTs7zNusWTN9/fXXWr58uWrUqKEWLVpo27ZtWa7L19dXGzduVMmSJdWlSxdVrlxZPXv2VEpKiv1D0a+++qq6d++uiIgINWjQQP7+/nr00Ucd5hkzZowiIyM1YcIEVa5cWW3bttW3336rMmXKZKmOu30eAAAAAJATXPquxTVq1NC0adPueI7WrVurcuXKmj59evYVlsckJiYqMDBQPq+9qKseLn0BgMuxurtrQZ0WemrnWj4jmwcY42fc1vi0tDRFR0erQ4cOfE7KBdF/10XvXRv9d21Z6f+NbMBdi53k4sWLWr16tdauXasZM27vlzsAAAAAwN0hyN6BWrVq6dKlS5o0aZLD52YBAAAAADnPZYPs+vXr7/jYuLi4bKsDAAAAAHB7XPpmTwAAAAAA8yHIAgAAAABMhSALAAAAADAVgiwAAAAAwFRc9mZPyF2nho1TcHCws8tALrrxXWIJke/wXXIAAADIVqzIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMxcPZBcA1lJg0Qlc9eLu5Equ7uxbUaaHAt4YoJT3d2eXgNhjjZzi7BAAAgFtiRRYAAAAAYCoEWQAAAACAqRBkAQAAAACmQpAFAAAAAJgKQRYAAAAAYCoEWQAAAACAqRBkAQAAAACmQpAFAAAAAJgKQRYOIiIi1LlzZ2eXAQAAAAA3RZAFAAAAAJgKQRYAAAAAYCoE2Vxis9k0adIklS9fXt7e3ipZsqTGjRsnSdq3b59atGghq9Wq4OBg9e7dW3/99Zf92BuX+44fP15FihRRUFCQRo8erevXr+u1115TgQIFVKJECc2cOdPhnKdPn1bXrl2VP39+BQcHq1OnToqLi7PvT09P1+DBgxUUFKTg4GANHTpUhmHY98+dO1fBwcFKTU11mPexxx5T9+7dc+BVAgAAAID/5uHsAlzF8OHD9emnn2rq1Kl68MEHdfbsWR06dEjJyclq166d6tevrx07dig+Pl4vvPCC+vXrp9mzZ9uPX7t2rUqUKKGNGzfqxx9/1PPPP6+tW7eqSZMm2rZtm7766iu99NJLat26tUJDQ5WcnKzmzZurcePG2rhxozw8PDR27Fi1a9dOe/fulZeXl6ZMmaKZM2fq888/V5UqVTRlyhQtXbpULVq0kCQ98cQTGjBggJYvX64nnnhCknT+/HmtWLFC33//fabPMzU11SH4JiYmSpKsbu6yuLvn0KuLe5HVzd3hb5hHWlpats2RHXPBfOi/66L3ro3+u7as9D873xsW459LcMgRV65cUaFChTRjxgy98MILDvs+/fRTDRs2TL///rv8/PwkSdHR0erYsaPOnDmjIkWKKCIiQuvXr9fx48fl5vb3InqlSpVUuHBhbdy4UdLfq6uBgYH67LPP1K1bN82cOVOTJ0/WwYMHZbFYJEnXrl1TUFCQli1bpjZt2qhYsWIaOHCghg0bJkm6fv26ypQpo9q1a2vZsmWSpD59+iguLk7R0dGSpPfee0/Tp0/X0aNH7fP+U1RUlEaPHp1h+/z58+Xr65sNryYAAAAAM0pOTtbTTz+thIQEBQQE3NVcrMjmgoMHDyo1NVUtW7bMdF/16tXtIVaSGjVqJJvNpsOHD6tIkSKSpPvuu88eYiWpSJEiuv/+++2P3d3dFRwcrPj4eEnSrl27dPToUfn7+zuc7+rVqzp27JgSEhJ09uxZNWjQwL7Pw8NDderUcbi8uFevXqpbt65Onz6t4sWLa9asWYqIiMg0xEp/rzwPHjzY/jgxMVGhoaHq+/NmXfXk7eZKrG7umlmrqXru3qAUW7qzy8FtSIh8567nSEtLU0xMjFq3bi1PT89sqApmQv9dF713bfTftWWl/zeu1swOJItcYLVab7rPMIybhsJ/bv/3m8FisWS6zWazSfr7M7m1a9fWvHnzMsxbqFChLNdes2ZNVa9eXXPnzlXbtm21b98+ffvttzcd7+3tLW9v7wzbU2zpupqe+fNE3pZiS1dKOkHWTLLzlw9PT09+mXFh9N910XvXRv9d2636n53vC272lAvCwsJktVq1Zs2aDPuqVKmi2NhYJSUl2bf9+OOPcnNzU4UKFe74nLVq1dKRI0dUuHBhlS9f3uFPYGCgAgMDFRISop9++sl+zPXr17Vr164Mc73wwguaNWuWZs6cqVatWik0NPSO6wIAAACAu0WQzQU+Pj4aNmyYhg4dqrlz5+rYsWP66aef9Pnnnys8PFw+Pj7q0aOHfvnlF61bt079+/fXs88+a7+s+E6Eh4erYMGC6tSpkzZt2qQTJ05ow4YNGjhwoE6dOiVJGjhwoCZOnKilS5fq0KFD6tOnjy5fvpzpXKdPn9ann36qnj173nFNAAAAAJAdCLK5ZOTIkXr11VcVGRmpypUrq2vXroqPj5evr69WrVqlixcvqm7dunr88cfVsmVLzZgx467O5+vrq40bN6pkyZLq0qWLKleurJ49eyolJcX+wepXX31V3bt3V0REhBo0aCB/f389+uijGeYKCAjQY489pnz58qlz5853VRcAAAAA3C0+I5tL3NzcNGLECI0YMSLDvqpVq2rt2rU3PfafX8Nzw/r16zNs++d3xEpS0aJFNWfOnJvO6+HhoWnTpmnatGk3HXPD2bNnFR4enunnXwEAAAAgNxFkcUsXL17U6tWrtXbt2rteJQYAAACA7ECQxS3VqlVLly5d0qRJk1SxYkVnlwMAAAAABFnc2r8vVwYAAAAAZ+NmTwAAAAAAUyHIAgAAAABMhSALAAAAADAVPiOLXHFq2DgFBwc7uwzkorS0NEVHRysh8h15eno6uxwAAADkIazIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIsgAAAAAAUyHIAgAAAABMhSALAAAAADAVD2cXANdQYtIIXfXg7eZKrO7uWlCnhQLfGqKU9PT/HG+Mn5ELVQEAACAvYEUWAAAAAGAqBFkAAAAAgKkQZAEAAAAApkKQBQAAAACYCkEWAAAAAGAqBFkAAAAAgKkQZAEAAAAApkKQBQAAAACYCkHWRf3xxx9q3bq1/Pz8FBQUlKVj1q9fL4vFosuXL+dobQAAAABwKx7OLgDOMXXqVJ09e1axsbEKDAx0djkAAAAAkGUEWRd17Ngx1a5dW2FhYc4uBQAAAABui8teWvzNN9+oatWqslqtCg4OVqtWrZSUlKRmzZpp0KBBDmM7d+6siIgI++MPPvhAYWFh8vHxUZEiRfT444/b9zVr1kz9+vVTv379FBQUpODgYL355psyDMM+5tq1axo6dKiKFy8uPz8/PfDAA1q/fr19/+zZsxUUFKRVq1apcuXKypcvn9q1a6ezZ8/ax6xfv1716tWzXxrcqFEj/fbbb/b93377rWrXri0fHx+VLVtWo0eP1vXr1yVJpUuX1uLFizV37lxZLBZFREQoLi5OFotFsbGx9jkuX74si8XiUBsAAAAAOJtLrsiePXtWTz31lCZPnqxHH31UV65c0aZNmxzC5s3s3LlTAwYM0BdffKGGDRvq4sWL2rRpk8OYOXPm6Pnnn9e2bdu0c+dO9e7dW6VKlVKvXr0kSc8995zi4uK0cOFCFStWTEuXLlW7du20b98++wppcnKy3nnnHX3xxRdyc3PTM888oyFDhmjevHm6fv26OnfurF69emnBggW6du2atm/fLovFIklatWqVnnnmGU2fPl2NGzfWsWPH1Lt3b0nSqFGjtGPHDnXv3l0BAQF67733ZLVadenSpWx5bVNTU5Wammp/nJiYKEmyurnL4u6eLeeAOVjd3B3+/i9paWk5WQ5y2Y1+0lfXRP9dF713bfTftWWl/9n53nDZIHv9+nV16dJFpUqVkiRVrVo1S8eePHlSfn5+evjhh+Xv769SpUqpZs2aDmNCQ0M1depUWSwWVaxYUfv27dPUqVPVq1cvHTt2TAsWLNCpU6dUrFgxSdKQIUP0/fffa9asWRo/frykv5v80UcfqVy5cpKkfv366a233pL0dzhMSEjQww8/bN9fuXJl+/nHjRun119/XT169JAklS1bVmPGjNHQoUM1atQoFSpUSN7e3rJarSpatKgkZVuQnTBhgkaPHp1h+/+qPyhfX99sOQfMZWatplkaFx0dncOVwBliYmKcXQKciP67Lnrv2ui/a7tV/5OTk7PtPC4ZZKtXr66WLVuqatWqatu2rdq0aaPHH39c+fPn/89jW7durVKlSqls2bJq166d2rVrp0cffdQhpNWvX9++OipJDRo00JQpU5Senq7du3fLMAxVqFDBYd7U1FQFBwfbH/v6+tpDqiSFhIQoPj5eklSgQAFFRESobdu2at26tVq1aqUnn3xSISEhkqRdu3Zpx44dGjdunP349PR0Xb16VcnJyTkaKIcPH67BgwfbHycmJio0NFR9f96sq54u+XZzWVY3d82s1VQ9d29Qii39P8cnRL6TC1Uht6SlpSkmJkatW7eWp6ens8tBLqP/roveuzb679qy0v8bV2tmB5dMFu7u7oqJidGWLVu0evVqvf/++xoxYoS2bdsmNze3DJcY/3MJ3N/fX7t379b69eu1evVqRUZGKioqSjt27MjS19jYbDa5u7tr165dcv/Xpbb58uWz//zv5lssFoe6Zs2apQEDBuj777/XV199pTfffFMxMTGqX7++bDabRo8erS5dumQ4v4+PT6Z1ubn9/XHpf57jTpb+vb295e3tnWF7ii1dV9MtmRyBvC7Flq6U9P8OsvyDlzd5enrSWxdG/10XvXdt9N+13ar/2fm+cNmbPVksFjVq1EijR4/Wnj175OXlpaVLl6pQoUION1VKT0/XL7/84nCsh4eHWrVqpcmTJ2vv3r2Ki4vT2rVr7ft/+uknh/E//fSTwsLC5O7urpo1ayo9PV3x8fEqX768w58bl/lmVc2aNTV8+HBt2bJF999/v+bPny9JqlWrlg4fPpxh/vLly9sD678VKlRIkhye+z9v/AQAAAAA9wqXXJHdtm2b1qxZozZt2qhw4cLatm2bzp07p8qVK8vPz0+DBw/WypUrVa5cOU2dOlWXL1+2H7tixQodP35cTZo0Uf78+RUdHS2bzaaKFSvax/z+++8aPHiwXnzxRe3evVvvv/++pkyZIkmqUKGCwsPD1b17d02ZMkU1a9bU+fPntXbtWlWtWlUdOnT4z/pPnDihTz75RI888oiKFSumw4cP69dff1X37t0lSZGRkXr44YcVGhqqJ554Qm5ubtq7d6/27dunsWPHZjqn1WpV/fr1NXHiRJUuXVrnz5/Xm2++eRevMgAAAADkDJcMsgEBAdq4caOmTZumxMRElSpVSlOmTFH79u2Vlpamn3/+Wd27d5eHh4deeeUVNW/e3H5sUFCQlixZoqioKF29elVhYWFasGCB7rvvPvuY7t27KyUlRfXq1ZO7u7v69+9vv2uw9PdlwWPHjtWrr76q06dPKzg4WA0aNMhSiJX+/vzsoUOHNGfOHF24cEEhISHq16+fXnzxRUlS27ZttWLFCr311luaPHmyPD09ValSJb3wwgu3nHfmzJnq2bOn6tSpo4oVK2ry5Mlq06bN7by0AAAAAJDjLEZWvnMGWdasWTPVqFFD06ZNc3Yp94TExEQFBgbK57UXddXDJf+7icuyurtrQZ0Wemrn2ix9RtYYPyMXqkJuSUtLU3R0tDp06MDnpFwQ/Xdd9N610X/XlpX+38gGCQkJCggIuKvzuexnZAEAAAAA5kSQBQAAAACYCtd6ZrP169c7uwQAAAAAyNNYkQUAAAAAmApBFgAAAABgKgRZAAAAAICp8BlZ5IpTw8YpODjY2WUgF924BXtC5Dvcgh8AAADZihVZAAAAAICpEGQBAAAAAKZCkAUAAAAAmApBFgAAAABgKgRZAAAAAICpEGQBAAAAAKZCkAUAAAAAmApBFgAAAABgKgRZAAAAAICpEGQBAAAAAKZCkAUAAAAAmApBFgAAAABgKgRZAAAAAICpEGQBAAAAAKZCkAUAAAAAmApBFgAAAABgKgRZAAAAAICpEGQBAAAAAKZCkAUAAAAAmApBFgAAAABgKgRZAAAAAICpEGQBAAAAAKZCkAUAAAAAmApBFgAAAABgKgRZAAAAAICpEGQBAAAAAKbi4ewC4BpKTBqhqx683czGGD/D2SUAAAAAGbAiCwAAAAAwFYIsAAAAAMBUCLIAAAAAAFMhyAIAAAAATIUgCwAAAAAwFYIsAAAAAMBUCLIAAAAAAFMhyAIAAAAATMVlg2zp0qU1bdo0Z5dhChaLRcuWLZMkxcXFyWKxKDY21qk1AQAAAHBdHs4uAPe+s2fPKn/+/M4uAwAAAAAkEWRz1bVr1+Tl5eXsMm5b0aJFnV0CAAAAANg59dLiK1euKDw8XH5+fgoJCdHUqVPVrFkzDRo0yD7m0qVL6t69u/Lnzy9fX1+1b99eR44ccZhn8eLFuu++++Tt7a3SpUtrypQpDvvj4+PVsWNHWa1WlSlTRvPmzbtlXatWrZKPj48uX77ssH3AgAFq2rSp/fGWLVvUpEkTWa1WhYaGasCAAUpKSrLvL126tMaOHauIiAgFBgaqV69eunbtmvr166eQkBD5+PiodOnSmjBhgqTML9u9fPmyLBaL1q9fb389wsPDVahQIVmtVoWFhWnWrFk3fS6GYWjy5MkqW7asrFarqlevrm+++UaSZLPZVKJECX300UcOx+zevVsWi0XHjx+X5HhpMQAAAAA4m1NXZAcPHqwff/xRy5cvV5EiRRQZGandu3erRo0a9jERERE6cuSIli9froCAAA0bNkwdOnTQgQMH5OnpqV27dunJJ59UVFSUunbtqi1btqhPnz4KDg5WRESEfY7ff/9da9eulZeXlwYMGKD4+Pib1tWqVSsFBQVp8eLFev755yVJ6enpWrRokd566y1J0r59+9S2bVuNGTNGn3/+uc6dO6d+/fqpX79+DsHy7bff1siRI/Xmm29KkqZPn67ly5dr0aJFKlmypH7//Xf9/vvvWX7NRo4cqQMHDui7775TwYIFdfToUaWkpNx0/JtvvqklS5boww8/VFhYmDZu3KhnnnlGhQoVUtOmTdWtWzfNmzdPL730kv2Y+fPnq0GDBipbtmyW67ohNTVVqamp9seJiYmSJKubuyzu7rc9H5wrLS3tro+9mzlgXvTftdF/10XvXRv9d21Z6X92vjcshmEY2Tbbbbhy5YqCg4M1f/58Pf7445KkhIQEFStWTL169dK0adN05MgRVahQQT/++KMaNmwoSbpw4YJCQ0M1Z84cPfHEEwoPD9e5c+e0evVq+9xDhw7VypUrtX//fv3666+qWLGifvrpJz3wwAOSpEOHDqly5cqaOnWqw+rvPw0cOFC//PKL1qxZI0lavXq1OnbsqD/++EP58+dX9+7dZbVa9fHHH9uP2bx5s5o2baqkpCT7amvNmjW1dOlS+5gBAwZo//79+uGHH2SxWBzOGRcXpzJlymjPnj32MH/58mXlz59f69atU7NmzfTII4+oYMGCmjlz5n++xklJSSpYsKDWrl2rBg0a2Le/8MILSk5O1vz587Vnzx7Vrl1bJ06cUKlSpWSz2VSyZEm98cYb6tOnj6S/V2SXLl2qzp07Z1rjP0VFRWn06NEZts+fP1++vr7/WTMAAACAvCk5OVlPP/20EhISFBAQcFdzOW1F9vjx40pLS1O9evXs2wIDA1WxYkX744MHD8rDw8MeQCUpODhYFStW1MGDB+1jOnXq5DB3o0aNNG3aNKWnp9vnqFOnjn1/pUqVFBQUdMv6wsPD1aBBA505c0bFihXTvHnz1KFDB/tNj3bt2qWjR486XKZsGIZsNptOnDihypUrS5LDeaW/V4dbt26tihUrql27dnr44YfVpk2brLxkkqSXX35Zjz32mHbv3q02bdqoc+fO9pD/bwcOHNDVq1fVunVrh+3Xrl1TzZo1JUk1a9ZUpUqVtGDBAr3++uvasGGD4uPj9eSTT2a5pn8aPny4Bg8ebH+cmJio0NBQ9f15s6568pFss0mIfOeOj01LS1NMTIxat24tT0/PbKwKZkD/XRv9d1303rXRf9eWlf7fuFozOzgtWdxYCP73quQ/F4hvtlhsGIb9uH/+fKs5/j3mv9SrV0/lypXTwoUL9fLLL2vp0qUOlwzbbDa9+OKLGjBgQIZjS5Ysaf/Zz8/PYV+tWrV04sQJfffdd/rhhx/05JNPqlWrVvrmm2/k5uaWof5/L7+3b99ev/32m1auXKkffvhBLVu2VN++ffXOOxkDh81mkyStXLlSxYsXd9jn7e1t/zk8PFzz58/X66+/rvnz56tt27YqWLDgf75GmfH29naY+4YUW7qupt9eD+B82fGPkKenJ/+YuTD679rov+ui966N/ru2W/U/O98XTrvZU7ly5eTp6ant27fbtyUmJjrcyKlKlSq6fv26tm3bZt924cIF/frrr/YVzypVqmjz5s0Oc2/ZskUVKlSQu7u7KleurOvXr2vnzp32/YcPH85wI6fMPP3005o3b56+/fZbubm56aGHHrLvq1Wrlvbv36/y5ctn+PNfdyYOCAhQ165d9emnn+qrr77S4sWLdfHiRRUqVEjS3193c0Nm39daqFAhRURE6Msvv9S0adP0ySefZHqeKlWqyNvbWydPnsxQY2hoqMPz3Ldvn3bt2qVvvvlG4eHh//naAAAAAICzOG1F1t/fXz169NBrr72mAgUKqHDhwho1apTc3Nzsq6dhYWHq1KmTevXqpY8//lj+/v56/fXXVbx4cfvlxK+++qrq1q2rMWPGqGvXrtq6datmzJihDz74QJLsl/D26tVLn3zyiTw8PDRo0CBZrdb/rDE8PFyjR4/WuHHj9Pjjj8vHx8e+b9iwYapfv7769u2rXr16yc/PTwcPHlRMTIzef//9m845depUhYSEqEaNGnJzc9PXX3+tokWLKigoSG5ubqpfv74mTpyo0qVL6/z58/abRN0QGRmp2rVr67777lNqaqpWrFhhD/WZvcZDhgzRK6+8IpvNpgcffFCJiYnasmWL8uXLpx49ekiSypQpo4YNG+r555/X9evXM1yqDQAAAAD3Eqd+/c67776rBg0a6OGHH1arVq3UqFEjVa5c2SEwzpo1S7Vr19bDDz+sBg0ayDAMRUdH25ela9WqpUWLFmnhwoW6//77FRkZqbfeest+x+Ibc4SGhqpp06bq0qWLevfurcKFC/9nfWFhYapbt6727t2bYZWyWrVq2rBhg44cOaLGjRurZs2aGjlypEJCQm45Z758+TRp0iTVqVNHdevWVVxcnKKjo+2XFc+cOVNpaWmqU6eOBg4cqLFjxzoc7+XlpeHDh6tatWpq0qSJ3N3dtXDhwpueb8yYMYqMjNSECRNUuXJltW3bVt9++63KlCnjMC48PFw///yzunTpkqWQDwAAAADO4rS7FmcmKSlJxYsX15QpU+xfewNzS0xMVGBgoHxee1FXPbjZk9kY42fc8bFpaWmKjo5Whw4d+JyMC6L/ro3+uy5679rov2vLSv9vZANT37VYkvbs2aNDhw6pXr16SkhIsH9HK5e2AgAAAABuxulLZO+8844OHz4sLy8v1a5dW5s2bbrjO+YCAAAAAPI+pwbZmjVrateuXc4sAQAAAABgMk692RMAAAAAALeLIAsAAAAAMBWCLAAAAADAVJx+sye4hlPDxik4ONjZZQAAAADIA1iRBQAAAACYCkEWAAAAAGAqBFkAAAAAgKkQZAEAAAAApkKQBQAAAACYCkEWAAAAAGAqBFkAAAAAgKkQZAEAAAAApkKQBQAAAACYCkEWAAAAAGAqBFkAAAAAgKkQZAEAAAAApuLh7AKQtxmGIUm6cuWKPD09nVwNclNaWpqSk5OVmJhI710Q/Xdt9N910XvXRv9dW1b6n5iYKOn/zwh3gyCLHHXhwgVJUpkyZZxcCQAAAIB7wZUrVxQYGHhXcxBkkaMKFCggSTp58uRdv1lhLomJiQoNDdXvv/+ugIAAZ5eDXEb/XRv9d1303rXRf9eWlf4bhqErV66oWLFid30+gixylJvb3x/DDgwM5P/QXFRAQAC9d2H037XRf9dF710b/Xdt/9X/7Frc4mZPAAAAAABTIcgCAAAAAEyFIIsc5e3trVGjRsnb29vZpSCX0XvXRv9dG/13XfTetdF/15bb/bcY2XHvYwAAAAAAcgkrsgAAAAAAUyHIAgAAAABMhSALAAAAADAVgiwAAAAAwFQIssgxH3zwgcqUKSMfHx/Vrl1bmzZtcnZJuE0bN25Ux44dVaxYMVksFi1btsxhv2EYioqKUrFixWS1WtWsWTPt37/fYUxqaqr69++vggULys/PT4888ohOnTrlMObSpUt69tlnFRgYqMDAQD377LO6fPlyDj873MqECRNUt25d+fv7q3DhwurcubMOHz7sMIb+510ffvihqlWrZv9S+wYNGui7776z76f3rmPChAmyWCwaNGiQfRv9z7uioqJksVgc/hQtWtS+n97nfadPn9Yzzzyj4OBg+fr6qkaNGtq1a5d9/z31HjCAHLBw4ULD09PT+PTTT40DBw4YAwcONPz8/IzffvvN2aXhNkRHRxsjRowwFi9ebEgyli5d6rB/4sSJhr+/v7F48WJj3759RteuXY2QkBAjMTHRPuall14yihcvbsTExBi7d+82mjdvblSvXt24fv26fUy7du2M+++/39iyZYuxZcsW4/777zcefvjh3HqayETbtm2NWbNmGb/88osRGxtrPPTQQ0bJkiWNv/76yz6G/uddy5cvN1auXGkcPnzYOHz4sPHGG28Ynp6exi+//GIYBr13Fdu3bzdKly5tVKtWzRg4cKB9O/3Pu0aNGmXcd999xtmzZ+1/4uPj7fvpfd528eJFo1SpUkZERISxbds248SJE8YPP/xgHD161D7mXnoPEGSRI+rVq2e89NJLDtsqVapkvP76606qCHfr30HWZrMZRYsWNSZOnGjfdvXqVSMwMND46KOPDMMwjMuXLxuenp7GwoUL7WNOnz5tuLm5Gd9//71hGIZx4MABQ5Lx008/2cds3brVkGQcOnQoh58Vsio+Pt6QZGzYsMEwDPrvivLnz2989tln9N5FXLlyxQgLCzNiYmKMpk2b2oMs/c/bRo0aZVSvXj3TffQ+7xs2bJjx4IMP3nT/vfYe4NJiZLtr165p165datOmjcP2Nm3aaMuWLU6qCtntxIkT+uOPPxz67O3traZNm9r7vGvXLqWlpTmMKVasmO6//377mK1btyowMFAPPPCAfUz9+vUVGBjI++UekpCQIEkqUKCAJPrvStLT07Vw4UIlJSWpQYMG9N5F9O3bVw899JBatWrlsJ3+531HjhxRsWLFVKZMGXXr1k3Hjx+XRO9dwfLly1WnTh098cQTKly4sGrWrKlPP/3Uvv9eew8QZJHtzp8/r/T0dBUpUsRhe5EiRfTHH384qSpktxu9vFWf//jjD3l5eSl//vy3HFO4cOEM8xcuXJj3yz3CMAwNHjxYDz74oO6//35J9N8V7Nu3T/ny5ZO3t7deeuklLV26VFWqVKH3LmDhwoXavXu3JkyYkGEf/c/bHnjgAc2dO1erVq3Sp59+qj/++EMNGzbUhQsX6L0LOH78uD788EOFhYVp1apVeumllzRgwADNnTtX0r33v3+PrD814PZYLBaHx4ZhZNgG87uTPv97TGbjeb/cO/r166e9e/dq8+bNGfbR/7yrYsWKio2N1eXLl7V48WL16NFDGzZssO+n93nT77//roEDB2r16tXy8fG56Tj6nze1b9/e/nPVqlXVoEEDlStXTnPmzFH9+vUl0fu8zGazqU6dOho/frwkqWbNmtq/f78+/PBDde/e3T7uXnkPsCKLbFewYEG5u7tn+C8q8fHxGf4LDszrxl0Mb9XnokWL6tq1a7p06dItx/z5558Z5j937hzvl3tA//79tXz5cq1bt04lSpSwb6f/eZ+Xl5fKly+vOnXqaMKECapevbree+89ep/H7dq1S/Hx8apdu7Y8PDzk4eGhDRs2aPr06fLw8LD3hv67Bj8/P1WtWlVHjhzhf/suICQkRFWqVHHYVrlyZZ08eVLSvfdvP0EW2c7Ly0u1a9dWTEyMw/aYmBg1bNjQSVUhu5UpU0ZFixZ16PO1a9e0YcMGe59r164tT09PhzFnz57VL7/8Yh/ToEEDJSQkaPv27fYx27ZtU0JCAu8XJzIMQ/369dOSJUu0du1alSlTxmE//Xc9hmEoNTWV3udxLVu21L59+xQbG2v/U6dOHYWHhys2NlZly5al/y4kNTVVBw8eVEhICP/bdwGNGjXK8FV7v/76q0qVKiXpHvy3P8u3hQJuw42v3/n888+NAwcOGIMGDTL8/PyMuLg4Z5eG23DlyhVjz549xp49ewxJxrvvvmvs2bPH/jVKEydONAIDA40lS5YY+/btM5566qlMb8FeokQJ44cffjB2795ttGjRItNbsFerVs3YunWrsXXrVqNq1archt/JXn75ZSMwMNBYv369w9cwJCcn28fQ/7xr+PDhxsaNG40TJ04Ye/fuNd544w3Dzc3NWL16tWEY9N7V/POuxYZB//OyV1991Vi/fr1x/Phx46effjIefvhhw9/f3/77G73P27Zv3254eHgY48aNM44cOWLMmzfP8PX1Nb788kv7mHvpPUCQRY753//+Z5QqVcrw8vIyatWqZf/aDpjHunXrDEkZ/vTo0cMwjL9vwz5q1CijaNGihre3t9GkSRNj3759DnOkpKQY/fr1MwoUKGBYrVbj4YcfNk6ePOkw5sKFC0Z4eLjh7+9v+Pv7G+Hh4calS5dy6VkiM5n1XZIxa9Ys+xj6n3f17NnT/v/fhQoVMlq2bGkPsYZB713Nv4Ms/c+7bnwnqKenp1GsWDGjS5cuxv79++376X3e9+233xr333+/4e3tbVSqVMn45JNPHPbfS+8Bi2EYRtbXbwEAAAAAcC4+IwsAAAAAMBWCLAAAAADAVAiyAAAAAABTIcgCAAAAAEyFIAsAAAAAMBWCLAAAAADAVAiyAAAAAABTIcgCAAAAAEyFIAsAAAAAMBWCLAAAuKk//vhD/fv3V9myZeXt7a3Q0FB17NhRa9asydU6LBaLli1blqvnBADcuzycXQAAALg3xcXFqVGjRgoKCtLkyZNVrVo1paWladWqVerbt68OHTrk7BIBAC7KYhiG4ewiAADAvadDhw7au3evDh8+LD8/P4d9ly9fVlBQkE6ePKn+/ftrzZo1cnNzU7t27fT++++rSJEikqSIiAhdvnzZYTV10KBBio2N1fr16yVJzZo1U7Vq1eTj46PPPvtMXl5eeumllxQVFSVJKl26tH777Tf78aVKlVJcXFxOPnUAwD2OS4sBAEAGFy9e1Pfff6++fftmCLGSFBQUJMMw1LlzZ128eFEbNmxQTEyMjh07pq5du972+ebMmSM/Pz9t27ZNkydP1ltvvaWYmBhJ0o4dOyRJs2bN0tmzZ+2PAQCui0uLAQBABkePHpVhGKpUqdJNx/zwww/au3evTpw4odDQUEnSF198ofvuu087duxQ3bp1s3y+atWqadSoUZKksLAwzZgxQ2vWrFHr1q1VqFAhSX+H56JFi97FswIA5BWsyAIAgAxufPLIYrHcdMzBgwcVGhpqD7GSVKVKFQUFBengwYO3db5q1ao5PA4JCVF8fPxtzQEAcB0EWQAAkEFYWJgsFsstA6lhGJkG3X9ud3Nz079vx5GWlpbhGE9PT4fHFotFNpvtTkoHALgAgiwAAMigQIECatu2rf73v/8pKSkpw/7Lly+rSpUqOnnypH7//Xf79gMHDighIUGVK1eWJBUqVEhnz551ODY2Nva26/H09FR6evptHwcAyJsIsgAAIFMffPCB0tPTVa9ePS1evFhHjhzRwYMHNX36dDVo0ECtWrVStWrVFB4ert27d2v79u3q3r27mjZtqjp16kiSWrRooZ07d2ru3Lk6cuSIRo0apV9++eW2ayldurTWrFmjP/74Q5cuXcrupwoAMBmCLAAAyFSZMmW0e/duNW/eXK+++qruv/9+tW7dWmvWrNGHH34oi8WiZcuWKX/+/GrSpIlatWqlsmXL6quvvrLP0bZtW40cOVJDhw5V3bp1deXKFXXv3v22a5kyZYpiYmIUGhqqmjVrZufTBACYEN8jCwAAAAAwFVZkAQAAAACmQpAFAAAAAJgKQRYAAAAAYCoEWQAAAACAqRBkAQAAAACmQpAFAAAAAJgKQRYAAAAAYCoEWQAAAACAqRBkAQAAAACmQpAFAAAAAJgKQRYAAAAAYCr/H/uu+nwEjXJfAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 1000x700 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "tags_pd.tail(10).plot(figsize=(10, 7),kind='barh', y='count', x='tags', grid=True, legend=False, color='#00796B')\n",
    "plt.xlabel('Count')\n",
    "plt.ylabel('Tag')\n",
    "plt.title('Top Ten Movie Tags')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "91cbc642-214c-49c6-ae03-280d0c728689",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['good versus evil',\n",
       " 'suspenseful',\n",
       " 'comedy',\n",
       " 'psychedelic',\n",
       " 'revenge',\n",
       " 'cult',\n",
       " 'romantic',\n",
       " 'flashback',\n",
       " 'violence',\n",
       " 'murder']"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "top_tags = tags_pd['tags'].tail(10).to_list()\n",
    "top_tags"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "68fd676f-83eb-43a6-8cf9-b6e8f90ea889",
   "metadata": {},
   "source": [
    "## 2) Machine Learning Models\n",
    "\n",
    "***a. Use a SparkML pipeline to build models to classify movies into the top 10 most popular tags.***"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "9632e3df-12aa-4c55-90b0-9716ea53f8c0",
   "metadata": {},
   "outputs": [],
   "source": [
    "from pyspark.ml import Pipeline\n",
    "from pyspark.ml.feature import Tokenizer, StopWordsRemover, CountVectorizer, HashingTF, IDF, VectorAssembler\n",
    "from pyspark.ml.classification import LogisticRegression"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "428b81ef-40c4-4312-83ee-5ab57efb9c1b",
   "metadata": {},
   "source": [
    "***b. The NLP pipeline should include removal of stopwords and tokenization of the syopsis data***\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "1045c720-3f64-43fb-9a9c-7cc19b762a13",
   "metadata": {},
   "outputs": [],
   "source": [
    "# tokenize synopsis data\n",
    "tokenizer = Tokenizer(inputCol=\"plot_synopsis\", outputCol=\"token_plot\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "783fe3da-da06-4d3e-a52e-634df9d8fd6c",
   "metadata": {},
   "outputs": [],
   "source": [
    "#remove stop words\n",
    "remover = StopWordsRemover(inputCol=\"token_plot\", outputCol=\"processed_plot\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "03599d51-1c6d-4ccf-9746-a74ff5bf972e",
   "metadata": {},
   "source": [
    "***c. The NLP pipeline should include more than one vectorization methods on the synopsis text for\n",
    "comparison (e.g. Count Vectorization, HashingTF, TF-IDF, BERT, etc.)***"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "38a5a7d6-fcc5-4493-8f6d-48d415821fce",
   "metadata": {},
   "outputs": [],
   "source": [
    "from pyspark.ml.feature import HashingTF, IDF, CountVectorizer\n",
    "\n",
    "# implementing HashingTF\n",
    "hashingTF = HashingTF(inputCol=\"processed_plot\", outputCol=\"rawFeatures\", numFeatures=2**18)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "872569b7-aef6-4133-aeb8-307200e4c52c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# implementing TF-IDF\n",
    "idf = IDF(inputCol=\"rawFeatures\", outputCol=\"tfidf_features\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "3fffdaf3-5eaf-41eb-85bb-7f8f04abc88a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# implementing count vectorizer\n",
    "count_vectorizer = CountVectorizer(inputCol=\"token_plot\", outputCol=\"count_vectorized\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "1284ae9f-724d-44e0-8379-51957a43d018",
   "metadata": {},
   "outputs": [],
   "source": [
    "# creating feature vector with multiple methods\n",
    "assembler = VectorAssembler(\n",
    "    inputCols=[\"rawFeatures\", \"tfidf_features\", \"count_vectorized\"],\n",
    "    outputCol=\"features\"\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "18e1e50a-f443-4afe-99d2-33fe52615011",
   "metadata": {},
   "source": [
    "***d. Pivot the data such that you are able to train one model for each genre tag.***\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "1fda66ab-4304-4d60-b966-162f4232fcc4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- title: string (nullable = true)\n",
      " |-- plot_synopsis: string (nullable = true)\n",
      " |-- split: string (nullable = true)\n",
      " |-- good versus evil: integer (nullable = false)\n",
      " |-- suspenseful: integer (nullable = false)\n",
      " |-- comedy: integer (nullable = false)\n",
      " |-- psychedelic: integer (nullable = false)\n",
      " |-- revenge: integer (nullable = false)\n",
      " |-- cult: integer (nullable = false)\n",
      " |-- romantic: integer (nullable = false)\n",
      " |-- flashback: integer (nullable = false)\n",
      " |-- violence: integer (nullable = false)\n",
      " |-- murder: integer (nullable = false)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "pivoted_df = (\n",
    "    df.groupBy(\"title\", \"plot_synopsis\", \"split\", \"tags\")  \n",
    "    .agg(*[(F.when(F.col(\"tags\").contains(tag), 1).otherwise(0)).alias(tag) for tag in top_tags])\n",
    ")\n",
    "pivoted_df = pivoted_df.drop('tags')\n",
    "pivoted_df.printSchema()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d1014a2f-ac89-4679-b1be-792f9ef4a29f",
   "metadata": {},
   "source": [
    "***e. The model should take in the movie synposis text as input features and individual tags as labels. Use the\n",
    "existing column “split” to split the data into train and test.***"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "f4a6a887-0406-4731-8825-0ebb04a81777",
   "metadata": {},
   "outputs": [],
   "source": [
    "orig_train = pivoted_df.filter(F.col('split') == 'train')\n",
    "orig_test = pivoted_df.filter(F.col('split') == 'test')\n",
    "orig_val = pivoted_df.filter(F.col('split') == 'val')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "f2f4be26-1907-410e-9300-1545736fef2c",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/27 19:28:31 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 55.1 MiB\n",
      "[Stage 81:>                                                         (0 + 1) / 1]\r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---------------+-----+----------------+-----------+------+-----------+-------+----+--------+---------+--------+------+--------------------+--------------------+--------------------+--------------------+--------------------+--------------------+\n",
      "|          title|split|good versus evil|suspenseful|comedy|psychedelic|revenge|cult|romantic|flashback|violence|murder|          token_plot|      processed_plot|         rawFeatures|      tfidf_features|    count_vectorized|            features|\n",
      "+---------------+-----+----------------+-----------+------+-----------+-------+----+--------+---------+--------+------+--------------------+--------------------+--------------------+--------------------+--------------------+--------------------+\n",
      "|           Nell|train|               0|          0|     0|          0|      0|   0|       0|        0|       0|     0|[when, stroke, vi...|[stroke, victim, ...|(262144,[1546,297...|(262144,[1546,297...|(262144,[0,1,2,3,...|(786432,[1546,297...|\n",
      "|        Get Low|train|               0|          0|     0|          0|      0|   0|       1|        0|       0|     0|[old, man, felix,...|[old, man, felix,...|(262144,[161,3388...|(262144,[161,3388...|(262144,[0,1,2,3,...|(786432,[161,3388...|\n",
      "|        Inferno|train|               0|          1|     0|          0|      0|   1|       0|        0|       1|     1|[rose, elliot, (i...|[rose, elliot, (i...|(262144,[301,2039...|(262144,[301,2039...|(262144,[0,1,2,3,...|(786432,[301,2039...|\n",
      "|Man of the Year|train|               0|          0|     1|          0|      0|   0|       0|        0|       0|     0|[tom, dobbs, (rob...|[tom, dobbs, (rob...|(262144,[75,238,6...|(262144,[75,238,6...|(262144,[0,1,2,3,...|(786432,[75,238,6...|\n",
      "|  49th Parallel|train|               0|          0|     0|          0|      0|   0|       0|        0|       0|     1|[u-37,, the, firs...|[u-37,, first, ge...|(262144,[180,1546...|(262144,[180,1546...|(262144,[0,1,2,3,...|(786432,[180,1546...|\n",
      "+---------------+-----+----------------+-----------+------+-----------+-------+----+--------+---------+--------+------+--------------------+--------------------+--------------------+--------------------+--------------------+--------------------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                                \r"
     ]
    }
   ],
   "source": [
    "# creating NLP pipeline\n",
    "\n",
    "pipeline = Pipeline(stages=[tokenizer, remover, hashingTF, idf, count_vectorizer, assembler])\n",
    "\n",
    "# fit the pipeline\n",
    "fitted_pipeline = pipeline.fit(orig_train)\n",
    "\n",
    "# transform data\n",
    "train = fitted_pipeline.transform(orig_train)\n",
    "train = train.drop('plot_synopsis')\n",
    "\n",
    "train.show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "b9d1eedc-0180-4bad-8729-26c421265bdb",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/27 19:28:43 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 55.1 MiB\n",
      "[Stage 84:>                                                         (0 + 1) / 1]\r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+--------------------+\n",
      "|            features|\n",
      "+--------------------+\n",
      "|(786432,[1546,297...|\n",
      "|(786432,[161,3388...|\n",
      "|(786432,[301,2039...|\n",
      "|(786432,[75,238,6...|\n",
      "|(786432,[180,1546...|\n",
      "+--------------------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                                \r"
     ]
    }
   ],
   "source": [
    "train.select('features').show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "c49085be-f393-4285-beb3-fbd7e4d03437",
   "metadata": {},
   "outputs": [],
   "source": [
    "# transform test and val with the pipeline\n",
    "\n",
    "test = fitted_pipeline.transform(orig_test)\n",
    "val = fitted_pipeline.transform(orig_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "6d983bd8-704c-4704-b537-a8169c09ec35",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/27 19:29:07 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 55.1 MiB\n",
      "[Stage 87:>                                                         (0 + 1) / 1]\r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---------------+----------------+-----------+------+-----------+-------+----+--------+---------+--------+------+--------------------+\n",
      "|          title|good versus evil|suspenseful|comedy|psychedelic|revenge|cult|romantic|flashback|violence|murder|            features|\n",
      "+---------------+----------------+-----------+------+-----------+-------+----+--------+---------+--------+------+--------------------+\n",
      "|           Nell|               0|          0|     0|          0|      0|   0|       0|        0|       0|     0|(786432,[1546,297...|\n",
      "|        Get Low|               0|          0|     0|          0|      0|   0|       1|        0|       0|     0|(786432,[161,3388...|\n",
      "|        Inferno|               0|          1|     0|          0|      0|   1|       0|        0|       1|     1|(786432,[301,2039...|\n",
      "|Man of the Year|               0|          0|     1|          0|      0|   0|       0|        0|       0|     0|(786432,[75,238,6...|\n",
      "|  49th Parallel|               0|          0|     0|          0|      0|   0|       0|        0|       0|     1|(786432,[180,1546...|\n",
      "+---------------+----------------+-----------+------+-----------+-------+----+--------+---------+--------+------+--------------------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                                \r"
     ]
    }
   ],
   "source": [
    "# dropping unnecessary columns\n",
    "cols_to_drop = ['split', 'plot_synopsis', 'token_plot', 'processed_plot', 'rawFeatures', 'tfidf_features', 'count_vectorized']\n",
    "\n",
    "train = train.drop(*cols_to_drop)\n",
    "test = test.drop(*cols_to_drop)\n",
    "val = val.drop(*cols_to_drop)\n",
    "\n",
    "train.show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "842e7eb5-69c5-4b1d-a682-3cd50e32e538",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Stage 88:>                                                         (0 + 1) / 1]\r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Partitions: 9\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Stage 89:>                                                         (0 + 1) / 1]\r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-----------+-----+\n",
      "|partitionId|count|\n",
      "+-----------+-----+\n",
      "|          8|  201|\n",
      "|          2| 1087|\n",
      "|          7| 1135|\n",
      "|          1| 1146|\n",
      "|          0| 1159|\n",
      "|          3| 1168|\n",
      "|          5| 1172|\n",
      "|          4| 1184|\n",
      "|          6| 1195|\n",
      "+-----------+-----+\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                                \r"
     ]
    }
   ],
   "source": [
    "#display number of records by partition\n",
    "def displaypartitions(df):\n",
    "    #number of records by partition\n",
    "    num = df.rdd.getNumPartitions()\n",
    "    print(\"Partitions:\", num)\n",
    "    df.withColumn(\"partitionId\", F.spark_partition_id())\\\n",
    "        .groupBy(\"partitionId\")\\\n",
    "        .count()\\\n",
    "        .orderBy(F.asc(\"count\"))\\\n",
    "        .show(num)\n",
    "\n",
    "displaypartitions(train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "4a10747e-3c5d-4e3c-808f-321ae0044f5f",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Stage 95:>                                                         (0 + 1) / 1]\r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Partitions: 9\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Stage 96:>                                                         (0 + 1) / 1]\r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-----------+-----+\n",
      "|partitionId|count|\n",
      "+-----------+-----+\n",
      "|          8|   85|\n",
      "|          0|  327|\n",
      "|          4|  339|\n",
      "|          1|  349|\n",
      "|          3|  356|\n",
      "|          7|  363|\n",
      "|          6|  367|\n",
      "|          2|  387|\n",
      "|          5|  392|\n",
      "+-----------+-----+\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                                \r"
     ]
    }
   ],
   "source": [
    "displaypartitions(test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "a9a32353-c212-4002-ba9b-cafae9fcc343",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- title: string (nullable = true)\n",
      " |-- good versus evil: integer (nullable = false)\n",
      " |-- suspenseful: integer (nullable = false)\n",
      " |-- comedy: integer (nullable = false)\n",
      " |-- psychedelic: integer (nullable = false)\n",
      " |-- revenge: integer (nullable = false)\n",
      " |-- cult: integer (nullable = false)\n",
      " |-- romantic: integer (nullable = false)\n",
      " |-- flashback: integer (nullable = false)\n",
      " |-- violence: integer (nullable = false)\n",
      " |-- murder: integer (nullable = false)\n",
      " |-- features: vector (nullable = true)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "train.printSchema()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "c7f93bbb-d09e-4734-bc62-7d98578c6fe1",
   "metadata": {},
   "outputs": [],
   "source": [
    "train = train.repartition(50).cache()\n",
    "test = test.repartition(50).cache()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f32cb0e9-ebe5-4134-9093-ea443dde552c",
   "metadata": {},
   "source": [
    "***f. If you use a One vs Rest classification approach to build your model, display validation scores against\n",
    "models for each of the 10 tags. Your final output should be a table with tag, f1 and accuracy. You should\n",
    "have 3 columns and 10 rows.***"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "85006930-9fec-4a03-9a1c-86078217aef4",
   "metadata": {},
   "outputs": [],
   "source": [
    "from pyspark.ml.classification import LogisticRegression, OneVsRest\n",
    "from pyspark.ml.evaluation import MulticlassClassificationEvaluator\n",
    "\n",
    "# setting up logistic regression as base classifier\n",
    "lr = LogisticRegression(maxIter=10, regParam=0.1)\n",
    "\n",
    "# instantiate one-vs-rest classifier using base classifier\n",
    "ovr = OneVsRest(classifier=lr, labelCol=f\"label\", featuresCol=\"features\", predictionCol=\"prediction\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "96fb999f-8f1d-4278-9adc-f2d5272f50a1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# create evaluators for F1 and accuracy scores\n",
    "\n",
    "evaluator_f1 = MulticlassClassificationEvaluator(metricName=\"f1\", labelCol=\"label\", predictionCol=\"prediction\")\n",
    "evaluator_accuracy = MulticlassClassificationEvaluator(metricName=\"accuracy\", labelCol=\"label\", predictionCol=\"prediction\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bfb5fc68-5f9e-4dc0-a7f4-78deaf28198f",
   "metadata": {},
   "source": [
    "**Note:** Output for the model execution cell is collapsed because it contains a lot of warnings about task size every time the model was executed so it is a lot of unecessary scrolling. I tried increasing executor memory, caching and suppressing warnings but it would not go away."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fad10acf-7a1d-46c5-a0b6-45aabe68f181",
   "metadata": {
    "collapsed": true,
    "jupyter": {
     "outputs_hidden": true
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluating model for tag: good versus evil\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/27 19:30:04 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 55.1 MiB\n",
      "23/11/27 19:31:25 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:31:35 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:31:41 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:31:47 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:31:55 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:32:00 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:32:06 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:32:11 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:32:17 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:32:21 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:32:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:32:32 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:32:39 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:32:43 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:32:49 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:32:54 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:32:59 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:33:04 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:33:10 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:33:15 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:33:20 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:33:25 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:33:32 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:33:36 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:33:43 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:33:47 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:33:54 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:33:58 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:34:05 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:34:09 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:34:15 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:34:20 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:34:29 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:34:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:34:40 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:34:45 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:34:50 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:34:55 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:35:00 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:35:05 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:35:11 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:35:15 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:35:21 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:35:25 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:35:31 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:35:36 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:35:42 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:35:46 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:35:53 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:35:57 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:36:04 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:36:08 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:36:17 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:36:22 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:36:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:36:32 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:36:38 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:36:42 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:36:48 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:36:52 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:36:58 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:37:17 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 55.1 MiB\n",
      "23/11/27 19:37:48 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "23/11/27 19:38:12 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "                                                                                \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluating model for tag: suspenseful\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/27 19:38:19 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:38:27 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:38:39 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:38:44 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:38:51 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:38:56 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:39:02 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:39:07 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:39:13 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:39:18 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:39:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:39:32 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:39:38 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:39:43 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:39:49 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:39:53 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:40:00 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:40:04 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:40:10 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:40:15 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:40:21 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:40:25 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:40:31 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:40:36 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:40:42 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:40:46 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:40:52 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:40:56 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:41:02 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:41:07 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:41:12 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:41:17 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:41:23 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:41:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:41:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:41:39 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:41:46 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:41:51 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:41:57 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:42:02 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:42:08 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:42:12 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:42:18 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:42:23 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:42:29 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:42:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:42:40 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:42:45 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:42:51 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:42:56 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:43:02 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:43:06 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:43:12 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:43:17 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:43:23 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:43:27 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:43:33 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:43:38 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:43:44 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:43:48 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:43:55 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:44:00 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:44:06 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:44:10 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:44:17 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:44:21 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:44:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:44:32 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:44:38 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:44:49 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "23/11/27 19:45:04 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "                                                                                \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluating model for tag: comedy\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/27 19:45:12 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:45:20 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:45:26 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:45:31 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:45:38 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:45:43 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:45:49 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:45:53 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:45:59 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:46:04 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:46:10 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:46:14 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:46:21 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:46:26 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:46:31 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:46:36 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:46:43 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:46:47 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:46:53 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:46:58 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:47:04 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:47:09 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:47:15 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:47:20 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:47:26 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:47:30 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:47:36 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:47:41 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:47:48 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:47:53 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:47:59 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:48:04 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:48:11 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:48:16 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:48:23 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:48:27 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:48:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:48:38 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:48:44 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:48:49 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:48:55 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:49:00 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:49:06 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:49:10 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:49:17 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:49:21 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:49:27 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:49:31 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:49:38 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:49:42 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:49:48 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:49:53 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:50:00 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:50:04 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:50:10 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:50:15 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:50:21 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:50:26 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:50:32 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:50:36 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:50:42 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:50:52 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "23/11/27 19:51:06 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "                                                                                \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluating model for tag: psychedelic\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/27 19:51:16 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:51:25 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:51:31 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:51:36 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:51:42 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:51:47 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:51:53 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:51:58 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:52:04 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:52:08 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:52:15 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:52:19 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:52:25 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:52:30 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:52:36 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:52:40 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:52:46 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:52:51 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:52:57 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:53:01 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:53:08 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:53:12 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:53:18 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:53:23 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:53:29 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:53:33 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:53:39 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:53:44 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:53:51 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:53:55 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:54:01 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:54:06 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:54:13 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:54:17 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:54:23 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:54:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:54:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:54:39 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:54:45 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:54:49 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:54:55 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:54:59 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:55:05 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:55:09 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:55:15 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:55:19 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:55:26 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:55:30 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:55:40 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:55:44 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:55:51 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:55:55 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:56:01 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:56:12 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "23/11/27 19:56:26 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "                                                                                \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluating model for tag: revenge\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/27 19:56:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:56:41 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:56:48 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:56:53 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:57:00 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:57:04 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:57:10 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:57:15 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:57:21 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:57:25 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:57:31 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:57:36 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:57:42 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:57:46 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:57:52 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:57:57 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:58:03 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:58:07 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:58:13 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:58:18 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:58:24 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:58:29 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:58:35 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:58:40 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:58:46 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:58:51 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:58:57 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:59:01 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:59:08 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 19:59:13 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:59:20 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:59:24 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:59:30 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:59:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:59:40 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:59:45 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:59:51 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 19:59:55 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:00:02 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:00:06 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:00:12 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:00:17 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:00:24 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:00:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:00:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:00:39 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:00:45 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:00:49 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:00:55 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:01:00 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:01:06 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:01:11 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:01:17 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:01:27 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "23/11/27 20:01:40 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "                                                                                \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluating model for tag: cult\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/27 20:01:46 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:01:54 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:02:00 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:02:05 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:02:12 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:02:16 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:02:22 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:02:26 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:02:32 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:02:37 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:02:43 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:02:47 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:02:54 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:02:58 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:03:04 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:03:09 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:03:15 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:03:20 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:03:26 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:03:31 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:03:37 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:03:41 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:03:47 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:03:52 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:03:52 ERROR org.apache.spark.network.client.TransportClient: Failed to send RPC RPC 7773521330692493131 to /10.128.0.173:48672: java.nio.channels.ClosedChannelException\n",
      "java.nio.channels.ClosedChannelException\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.newClosedChannelException(AbstractChannel.java:957)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.write(AbstractChannel.java:865)\n",
      "\tat io.netty.channel.DefaultChannelPipeline$HeadContext.write(DefaultChannelPipeline.java:1367)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWrite0(AbstractChannelHandlerContext.java:717)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWriteAndFlush(AbstractChannelHandlerContext.java:764)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext$WriteTask.run(AbstractChannelHandlerContext.java:1071)\n",
      "\tat io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472)\n",
      "\tat io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\n",
      "\tat io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\n",
      "\tat io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\n",
      "\tat java.lang.Thread.run(Thread.java:750)\n",
      "23/11/27 20:03:52 ERROR org.apache.spark.network.client.TransportClient: Failed to send RPC RPC 7613846154787198533 to /10.128.0.173:48672: java.nio.channels.ClosedChannelException\n",
      "java.nio.channels.ClosedChannelException\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.newClosedChannelException(AbstractChannel.java:957)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.write(AbstractChannel.java:865)\n",
      "\tat io.netty.channel.DefaultChannelPipeline$HeadContext.write(DefaultChannelPipeline.java:1367)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWrite0(AbstractChannelHandlerContext.java:717)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWriteAndFlush(AbstractChannelHandlerContext.java:764)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext$WriteTask.run(AbstractChannelHandlerContext.java:1071)\n",
      "\tat io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472)\n",
      "\tat io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\n",
      "\tat io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\n",
      "\tat io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\n",
      "\tat java.lang.Thread.run(Thread.java:750)\n",
      "23/11/27 20:03:52 WARN org.apache.spark.storage.BlockManagerMasterEndpoint: Error trying to remove shuffle 210 from block manager BlockManagerId(34, hub-msca-bdp-dphub-students-test-ridhi-sw-b0pt.c.msca-bdp-student-ap.internal, 33985, None)\n",
      "java.io.IOException: Failed to send RPC RPC 7773521330692493131 to /10.128.0.173:48672: java.nio.channels.ClosedChannelException\n",
      "\tat org.apache.spark.network.client.TransportClient$RpcChannelListener.handleFailure(TransportClient.java:363)\n",
      "\tat org.apache.spark.network.client.TransportClient$StdChannelListener.operationComplete(TransportClient.java:340)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListener0(DefaultPromise.java:577)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListenersNow(DefaultPromise.java:551)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListeners(DefaultPromise.java:490)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.setValue0(DefaultPromise.java:615)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.setFailure0(DefaultPromise.java:608)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.tryFailure(DefaultPromise.java:117)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.safeSetFailure(AbstractChannel.java:993)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.write(AbstractChannel.java:865)\n",
      "\tat io.netty.channel.DefaultChannelPipeline$HeadContext.write(DefaultChannelPipeline.java:1367)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWrite0(AbstractChannelHandlerContext.java:717)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWriteAndFlush(AbstractChannelHandlerContext.java:764)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext$WriteTask.run(AbstractChannelHandlerContext.java:1071)\n",
      "\tat io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472)\n",
      "\tat io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\n",
      "\tat io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\n",
      "\tat io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\n",
      "\tat java.lang.Thread.run(Thread.java:750)\n",
      "Caused by: java.nio.channels.ClosedChannelException\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.newClosedChannelException(AbstractChannel.java:957)\n",
      "\t... 12 more\n",
      "23/11/27 20:03:52 WARN org.apache.spark.storage.BlockManagerMasterEndpoint: Error trying to remove broadcast 598 from block manager BlockManagerId(34, hub-msca-bdp-dphub-students-test-ridhi-sw-b0pt.c.msca-bdp-student-ap.internal, 33985, None)\n",
      "java.io.IOException: Failed to send RPC RPC 7613846154787198533 to /10.128.0.173:48672: java.nio.channels.ClosedChannelException\n",
      "\tat org.apache.spark.network.client.TransportClient$RpcChannelListener.handleFailure(TransportClient.java:363)\n",
      "\tat org.apache.spark.network.client.TransportClient$StdChannelListener.operationComplete(TransportClient.java:340)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListener0(DefaultPromise.java:577)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListenersNow(DefaultPromise.java:551)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListeners(DefaultPromise.java:490)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.setValue0(DefaultPromise.java:615)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.setFailure0(DefaultPromise.java:608)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.tryFailure(DefaultPromise.java:117)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.safeSetFailure(AbstractChannel.java:993)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.write(AbstractChannel.java:865)\n",
      "\tat io.netty.channel.DefaultChannelPipeline$HeadContext.write(DefaultChannelPipeline.java:1367)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWrite0(AbstractChannelHandlerContext.java:717)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWriteAndFlush(AbstractChannelHandlerContext.java:764)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext$WriteTask.run(AbstractChannelHandlerContext.java:1071)\n",
      "\tat io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472)\n",
      "\tat io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\n",
      "\tat io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\n",
      "\tat io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\n",
      "\tat java.lang.Thread.run(Thread.java:750)\n",
      "Caused by: java.nio.channels.ClosedChannelException\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.newClosedChannelException(AbstractChannel.java:957)\n",
      "\t... 12 more\n",
      "23/11/27 20:03:52 ERROR org.apache.spark.network.client.TransportClient: Failed to send RPC RPC 5021862301703655991 to /10.128.0.173:48672: java.nio.channels.ClosedChannelException\n",
      "java.nio.channels.ClosedChannelException\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.newClosedChannelException(AbstractChannel.java:957)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.write(AbstractChannel.java:865)\n",
      "\tat io.netty.channel.DefaultChannelPipeline$HeadContext.write(DefaultChannelPipeline.java:1367)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWrite0(AbstractChannelHandlerContext.java:717)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWriteAndFlush(AbstractChannelHandlerContext.java:764)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext$WriteTask.run(AbstractChannelHandlerContext.java:1071)\n",
      "\tat io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472)\n",
      "\tat io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\n",
      "\tat io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\n",
      "\tat io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\n",
      "\tat java.lang.Thread.run(Thread.java:750)\n",
      "23/11/27 20:03:52 WARN org.apache.spark.storage.BlockManagerMasterEndpoint: Error trying to remove broadcast 594 from block manager BlockManagerId(34, hub-msca-bdp-dphub-students-test-ridhi-sw-b0pt.c.msca-bdp-student-ap.internal, 33985, None)\n",
      "java.io.IOException: Failed to send RPC RPC 5021862301703655991 to /10.128.0.173:48672: java.nio.channels.ClosedChannelException\n",
      "\tat org.apache.spark.network.client.TransportClient$RpcChannelListener.handleFailure(TransportClient.java:363)\n",
      "\tat org.apache.spark.network.client.TransportClient$StdChannelListener.operationComplete(TransportClient.java:340)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListener0(DefaultPromise.java:577)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListenersNow(DefaultPromise.java:551)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListeners(DefaultPromise.java:490)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.setValue0(DefaultPromise.java:615)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.setFailure0(DefaultPromise.java:608)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.tryFailure(DefaultPromise.java:117)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.safeSetFailure(AbstractChannel.java:993)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.write(AbstractChannel.java:865)\n",
      "\tat io.netty.channel.DefaultChannelPipeline$HeadContext.write(DefaultChannelPipeline.java:1367)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWrite0(AbstractChannelHandlerContext.java:717)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWriteAndFlush(AbstractChannelHandlerContext.java:764)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext$WriteTask.run(AbstractChannelHandlerContext.java:1071)\n",
      "\tat io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472)\n",
      "\tat io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\n",
      "\tat io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\n",
      "\tat io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\n",
      "\tat java.lang.Thread.run(Thread.java:750)\n",
      "Caused by: java.nio.channels.ClosedChannelException\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.newClosedChannelException(AbstractChannel.java:957)\n",
      "\t... 12 more\n",
      "23/11/27 20:03:52 ERROR org.apache.spark.network.client.TransportClient: Failed to send RPC RPC 6860341044503499284 to /10.128.0.173:48672: java.nio.channels.ClosedChannelException\n",
      "java.nio.channels.ClosedChannelException\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.newClosedChannelException(AbstractChannel.java:957)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.write(AbstractChannel.java:865)\n",
      "\tat io.netty.channel.DefaultChannelPipeline$HeadContext.write(DefaultChannelPipeline.java:1367)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWrite0(AbstractChannelHandlerContext.java:717)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWriteAndFlush(AbstractChannelHandlerContext.java:764)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext$WriteTask.run(AbstractChannelHandlerContext.java:1071)\n",
      "\tat io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472)\n",
      "\tat io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\n",
      "\tat io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\n",
      "\tat io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\n",
      "\tat java.lang.Thread.run(Thread.java:750)\n",
      "23/11/27 20:03:52 WARN org.apache.spark.storage.BlockManagerMasterEndpoint: Error trying to remove shuffle 209 from block manager BlockManagerId(34, hub-msca-bdp-dphub-students-test-ridhi-sw-b0pt.c.msca-bdp-student-ap.internal, 33985, None)\n",
      "java.io.IOException: Failed to send RPC RPC 6860341044503499284 to /10.128.0.173:48672: java.nio.channels.ClosedChannelException\n",
      "\tat org.apache.spark.network.client.TransportClient$RpcChannelListener.handleFailure(TransportClient.java:363)\n",
      "\tat org.apache.spark.network.client.TransportClient$StdChannelListener.operationComplete(TransportClient.java:340)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListener0(DefaultPromise.java:577)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListenersNow(DefaultPromise.java:551)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListeners(DefaultPromise.java:490)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.setValue0(DefaultPromise.java:615)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.setFailure0(DefaultPromise.java:608)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.tryFailure(DefaultPromise.java:117)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.safeSetFailure(AbstractChannel.java:993)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.write(AbstractChannel.java:865)\n",
      "\tat io.netty.channel.DefaultChannelPipeline$HeadContext.write(DefaultChannelPipeline.java:1367)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWrite0(AbstractChannelHandlerContext.java:717)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWriteAndFlush(AbstractChannelHandlerContext.java:764)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext$WriteTask.run(AbstractChannelHandlerContext.java:1071)\n",
      "\tat io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472)\n",
      "\tat io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\n",
      "\tat io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\n",
      "\tat io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\n",
      "\tat java.lang.Thread.run(Thread.java:750)\n",
      "Caused by: java.nio.channels.ClosedChannelException\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.newClosedChannelException(AbstractChannel.java:957)\n",
      "\t... 12 more\n",
      "23/11/27 20:03:52 ERROR org.apache.spark.network.client.TransportClient: Failed to send RPC RPC 5966723002020992586 to /10.128.0.173:48672: java.nio.channels.ClosedChannelException\n",
      "java.nio.channels.ClosedChannelException\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.newClosedChannelException(AbstractChannel.java:957)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.write(AbstractChannel.java:865)\n",
      "\tat io.netty.channel.DefaultChannelPipeline$HeadContext.write(DefaultChannelPipeline.java:1367)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWrite0(AbstractChannelHandlerContext.java:717)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWriteAndFlush(AbstractChannelHandlerContext.java:764)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext$WriteTask.run(AbstractChannelHandlerContext.java:1071)\n",
      "\tat io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472)\n",
      "\tat io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\n",
      "\tat io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\n",
      "\tat io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\n",
      "\tat java.lang.Thread.run(Thread.java:750)\n",
      "23/11/27 20:03:52 WARN org.apache.spark.storage.BlockManagerMasterEndpoint: Error trying to remove broadcast 597 from block manager BlockManagerId(34, hub-msca-bdp-dphub-students-test-ridhi-sw-b0pt.c.msca-bdp-student-ap.internal, 33985, None)\n",
      "java.io.IOException: Failed to send RPC RPC 5966723002020992586 to /10.128.0.173:48672: java.nio.channels.ClosedChannelException\n",
      "\tat org.apache.spark.network.client.TransportClient$RpcChannelListener.handleFailure(TransportClient.java:363)\n",
      "\tat org.apache.spark.network.client.TransportClient$StdChannelListener.operationComplete(TransportClient.java:340)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListener0(DefaultPromise.java:577)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListenersNow(DefaultPromise.java:551)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListeners(DefaultPromise.java:490)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.setValue0(DefaultPromise.java:615)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.setFailure0(DefaultPromise.java:608)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.tryFailure(DefaultPromise.java:117)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.safeSetFailure(AbstractChannel.java:993)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.write(AbstractChannel.java:865)\n",
      "\tat io.netty.channel.DefaultChannelPipeline$HeadContext.write(DefaultChannelPipeline.java:1367)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWrite0(AbstractChannelHandlerContext.java:717)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWriteAndFlush(AbstractChannelHandlerContext.java:764)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext$WriteTask.run(AbstractChannelHandlerContext.java:1071)\n",
      "\tat io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472)\n",
      "\tat io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\n",
      "\tat io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\n",
      "\tat io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\n",
      "\tat java.lang.Thread.run(Thread.java:750)\n",
      "Caused by: java.nio.channels.ClosedChannelException\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.newClosedChannelException(AbstractChannel.java:957)\n",
      "\t... 12 more\n",
      "23/11/27 20:03:52 ERROR org.apache.spark.network.client.TransportClient: Failed to send RPC RPC 7664465045990368214 to /10.128.0.173:48672: java.nio.channels.ClosedChannelException\n",
      "java.nio.channels.ClosedChannelException\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.newClosedChannelException(AbstractChannel.java:957)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.write(AbstractChannel.java:865)\n",
      "\tat io.netty.channel.DefaultChannelPipeline$HeadContext.write(DefaultChannelPipeline.java:1367)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWrite0(AbstractChannelHandlerContext.java:717)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWriteAndFlush(AbstractChannelHandlerContext.java:764)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext$WriteTask.run(AbstractChannelHandlerContext.java:1071)\n",
      "\tat io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472)\n",
      "\tat io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\n",
      "\tat io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\n",
      "\tat io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\n",
      "\tat java.lang.Thread.run(Thread.java:750)\n",
      "23/11/27 20:03:52 WARN org.apache.spark.storage.BlockManagerMasterEndpoint: Error trying to remove broadcast 595 from block manager BlockManagerId(34, hub-msca-bdp-dphub-students-test-ridhi-sw-b0pt.c.msca-bdp-student-ap.internal, 33985, None)\n",
      "java.io.IOException: Failed to send RPC RPC 7664465045990368214 to /10.128.0.173:48672: java.nio.channels.ClosedChannelException\n",
      "\tat org.apache.spark.network.client.TransportClient$RpcChannelListener.handleFailure(TransportClient.java:363)\n",
      "\tat org.apache.spark.network.client.TransportClient$StdChannelListener.operationComplete(TransportClient.java:340)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListener0(DefaultPromise.java:577)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListenersNow(DefaultPromise.java:551)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListeners(DefaultPromise.java:490)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.setValue0(DefaultPromise.java:615)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.setFailure0(DefaultPromise.java:608)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.tryFailure(DefaultPromise.java:117)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.safeSetFailure(AbstractChannel.java:993)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.write(AbstractChannel.java:865)\n",
      "\tat io.netty.channel.DefaultChannelPipeline$HeadContext.write(DefaultChannelPipeline.java:1367)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWrite0(AbstractChannelHandlerContext.java:717)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWriteAndFlush(AbstractChannelHandlerContext.java:764)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext$WriteTask.run(AbstractChannelHandlerContext.java:1071)\n",
      "\tat io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472)\n",
      "\tat io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\n",
      "\tat io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\n",
      "\tat io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\n",
      "\tat java.lang.Thread.run(Thread.java:750)\n",
      "Caused by: java.nio.channels.ClosedChannelException\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.newClosedChannelException(AbstractChannel.java:957)\n",
      "\t... 12 more\n",
      "23/11/27 20:03:52 ERROR org.apache.spark.network.client.TransportClient: Failed to send RPC RPC 6089465903498320541 to /10.128.0.173:48672: java.nio.channels.ClosedChannelException\n",
      "java.nio.channels.ClosedChannelException\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.newClosedChannelException(AbstractChannel.java:957)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.write(AbstractChannel.java:865)\n",
      "\tat io.netty.channel.DefaultChannelPipeline$HeadContext.write(DefaultChannelPipeline.java:1367)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWrite0(AbstractChannelHandlerContext.java:717)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWriteAndFlush(AbstractChannelHandlerContext.java:764)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext$WriteTask.run(AbstractChannelHandlerContext.java:1071)\n",
      "\tat io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472)\n",
      "\tat io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\n",
      "\tat io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\n",
      "\tat io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\n",
      "\tat java.lang.Thread.run(Thread.java:750)\n",
      "23/11/27 20:03:52 WARN org.apache.spark.storage.BlockManagerMasterEndpoint: Error trying to remove broadcast 599 from block manager BlockManagerId(34, hub-msca-bdp-dphub-students-test-ridhi-sw-b0pt.c.msca-bdp-student-ap.internal, 33985, None)\n",
      "java.io.IOException: Failed to send RPC RPC 6089465903498320541 to /10.128.0.173:48672: java.nio.channels.ClosedChannelException\n",
      "\tat org.apache.spark.network.client.TransportClient$RpcChannelListener.handleFailure(TransportClient.java:363)\n",
      "\tat org.apache.spark.network.client.TransportClient$StdChannelListener.operationComplete(TransportClient.java:340)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListener0(DefaultPromise.java:577)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListenersNow(DefaultPromise.java:551)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListeners(DefaultPromise.java:490)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.setValue0(DefaultPromise.java:615)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.setFailure0(DefaultPromise.java:608)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.tryFailure(DefaultPromise.java:117)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.safeSetFailure(AbstractChannel.java:993)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.write(AbstractChannel.java:865)\n",
      "\tat io.netty.channel.DefaultChannelPipeline$HeadContext.write(DefaultChannelPipeline.java:1367)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWrite0(AbstractChannelHandlerContext.java:717)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWriteAndFlush(AbstractChannelHandlerContext.java:764)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext$WriteTask.run(AbstractChannelHandlerContext.java:1071)\n",
      "\tat io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472)\n",
      "\tat io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\n",
      "\tat io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\n",
      "\tat io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\n",
      "\tat java.lang.Thread.run(Thread.java:750)\n",
      "Caused by: java.nio.channels.ClosedChannelException\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.newClosedChannelException(AbstractChannel.java:957)\n",
      "\t... 12 more\n",
      "23/11/27 20:03:52 ERROR org.apache.spark.network.client.TransportClient: Failed to send RPC RPC 8859756755102880374 to /10.128.0.173:48672: java.nio.channels.ClosedChannelException\n",
      "java.nio.channels.ClosedChannelException\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.newClosedChannelException(AbstractChannel.java:957)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.write(AbstractChannel.java:865)\n",
      "\tat io.netty.channel.DefaultChannelPipeline$HeadContext.write(DefaultChannelPipeline.java:1367)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWrite0(AbstractChannelHandlerContext.java:717)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWriteAndFlush(AbstractChannelHandlerContext.java:764)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext$WriteTask.run(AbstractChannelHandlerContext.java:1071)\n",
      "\tat io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472)\n",
      "\tat io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\n",
      "\tat io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\n",
      "\tat io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\n",
      "\tat java.lang.Thread.run(Thread.java:750)\n",
      "23/11/27 20:03:52 WARN org.apache.spark.storage.BlockManagerMasterEndpoint: Error trying to remove broadcast 596 from block manager BlockManagerId(34, hub-msca-bdp-dphub-students-test-ridhi-sw-b0pt.c.msca-bdp-student-ap.internal, 33985, None)\n",
      "java.io.IOException: Failed to send RPC RPC 8859756755102880374 to /10.128.0.173:48672: java.nio.channels.ClosedChannelException\n",
      "\tat org.apache.spark.network.client.TransportClient$RpcChannelListener.handleFailure(TransportClient.java:363)\n",
      "\tat org.apache.spark.network.client.TransportClient$StdChannelListener.operationComplete(TransportClient.java:340)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListener0(DefaultPromise.java:577)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListenersNow(DefaultPromise.java:551)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.notifyListeners(DefaultPromise.java:490)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.setValue0(DefaultPromise.java:615)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.setFailure0(DefaultPromise.java:608)\n",
      "\tat io.netty.util.concurrent.DefaultPromise.tryFailure(DefaultPromise.java:117)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.safeSetFailure(AbstractChannel.java:993)\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.write(AbstractChannel.java:865)\n",
      "\tat io.netty.channel.DefaultChannelPipeline$HeadContext.write(DefaultChannelPipeline.java:1367)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWrite0(AbstractChannelHandlerContext.java:717)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext.invokeWriteAndFlush(AbstractChannelHandlerContext.java:764)\n",
      "\tat io.netty.channel.AbstractChannelHandlerContext$WriteTask.run(AbstractChannelHandlerContext.java:1071)\n",
      "\tat io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472)\n",
      "\tat io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:500)\n",
      "\tat io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\n",
      "\tat io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\n",
      "\tat io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\n",
      "\tat java.lang.Thread.run(Thread.java:750)\n",
      "Caused by: java.nio.channels.ClosedChannelException\n",
      "\tat io.netty.channel.AbstractChannel$AbstractUnsafe.newClosedChannelException(AbstractChannel.java:957)\n",
      "\t... 12 more\n",
      "23/11/27 20:03:59 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:04:03 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:04:09 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:04:13 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:04:20 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:04:24 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:04:31 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:04:35 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:04:42 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:04:46 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:04:52 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:04:57 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:05:03 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:05:07 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:05:13 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:05:18 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:05:24 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:05:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:05:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:05:38 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:05:44 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:05:49 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:05:55 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:05:59 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:06:05 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:06:10 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:06:16 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:06:21 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:06:27 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:06:31 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:06:38 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:06:42 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:06:48 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:06:58 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "23/11/27 20:07:13 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "                                                                                \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluating model for tag: romantic\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/27 20:07:18 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:07:25 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:07:31 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:07:36 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:07:43 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:07:47 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:07:53 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:07:57 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:08:03 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:08:08 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:08:13 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:08:18 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:08:25 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:08:29 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:08:35 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:08:39 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:08:46 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:08:50 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:08:56 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:09:01 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:09:07 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:09:12 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:09:18 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:09:22 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:09:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:09:32 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:09:38 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:09:43 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:09:49 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:09:54 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:10:01 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:10:06 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:10:12 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:10:16 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:10:23 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:10:27 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:10:33 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:10:37 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:10:43 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:10:47 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:10:53 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:10:57 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:11:02 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:11:07 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:11:13 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:11:17 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:11:23 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:11:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:11:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:11:39 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:11:45 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:11:50 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:11:56 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:12:07 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "23/11/27 20:12:21 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "                                                                                \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluating model for tag: flashback\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/27 20:12:26 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:12:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:12:40 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:12:45 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:12:51 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:12:56 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:13:02 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:13:06 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:13:13 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:13:17 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:13:23 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:13:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:13:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:13:39 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:13:45 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:13:49 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:13:56 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:14:00 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:14:06 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:14:11 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:14:17 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:14:21 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:14:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:14:33 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:14:39 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:14:44 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:14:50 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:14:54 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:15:01 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:15:05 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:15:12 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:15:17 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:15:23 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:15:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:15:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:15:39 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:15:45 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:15:49 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:15:55 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:15:59 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:16:06 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:16:10 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:16:16 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:16:20 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:16:27 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:16:31 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:16:37 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:16:41 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:16:48 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:16:52 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:16:58 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:17:02 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:17:08 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:17:13 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:17:19 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:17:23 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:17:29 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:17:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:17:40 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:17:44 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:17:51 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:18:01 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "23/11/27 20:18:15 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "                                                                                \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluating model for tag: violence\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/27 20:18:21 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:18:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:18:35 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:18:40 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:18:46 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:18:50 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:18:56 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:19:01 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:19:07 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:19:11 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:19:17 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:19:22 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:19:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:19:32 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:19:43 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:19:49 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:19:54 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:20:00 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:20:04 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:20:10 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:20:15 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:20:21 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:20:26 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:20:32 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:20:36 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:20:43 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:20:47 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:20:53 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:20:58 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:21:05 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:21:10 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:21:16 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:21:21 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:21:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:21:32 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:21:39 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:21:43 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:21:57 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:22:01 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:22:08 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:22:12 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:22:19 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:22:24 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:22:30 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:22:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:22:41 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:22:45 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:22:51 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:22:55 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:23:02 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:23:06 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:23:12 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:23:16 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:23:23 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:23:27 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:23:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:23:45 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "23/11/27 20:23:59 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "                                                                                \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluating model for tag: murder\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/27 20:24:05 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:24:12 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:24:18 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:24:23 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:24:29 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:24:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:24:40 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:24:44 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:24:50 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:24:55 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:25:01 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:25:05 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:25:12 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:25:16 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:25:23 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:25:27 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:25:33 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:25:37 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:25:44 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:25:48 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:25:55 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:25:59 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:26:06 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:26:10 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:26:17 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:26:21 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:26:32 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:26:37 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:26:43 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "23/11/27 20:26:48 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:26:58 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:27:03 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:27:08 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:27:13 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:27:19 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:27:23 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:27:30 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:27:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:27:41 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:27:45 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:27:52 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:27:56 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:28:03 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:28:07 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:28:13 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:28:18 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:28:24 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:28:29 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:28:36 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:28:40 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:28:46 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:28:51 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:28:57 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.3 MiB\n",
      "23/11/27 20:29:08 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "23/11/27 20:29:21 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 60.3 MiB\n",
      "                                                                                \r"
     ]
    }
   ],
   "source": [
    "import time\n",
    "\n",
    "results = []\n",
    "tag_results = {}\n",
    "models = []\n",
    "\n",
    "start_time = time.time()\n",
    "\n",
    "# Loop through each tag and train a separate model while testing pre\n",
    "for tag in top_tags:\n",
    "    print(f\"Evaluating model for tag: {tag}\")\n",
    "\n",
    "    # Rename columns for evaluation\n",
    "    eval_train = train.withColumnRenamed(tag, \"label\")\n",
    "    eval_test = test.withColumnRenamed(tag, \"label\")\n",
    "    eval_train = eval_train.cache()\n",
    "    eval_test = eval_test.cache()\n",
    "    \n",
    "    # Fit the OneVsRest model\n",
    "    ovr_model = ovr.fit(eval_train)\n",
    "\n",
    "    # Make predictions on the test data\n",
    "    predictions = ovr_model.transform(eval_test)\n",
    "\n",
    "    # Store F1 score in the dictionary\n",
    "    f1_score = evaluator_f1.evaluate(predictions)\n",
    "\n",
    "    # Store accuracy in the dictionary\n",
    "    accuracy_score = evaluator_accuracy.evaluate(predictions)\n",
    "\n",
    "    results.append((tag, f1_score, accuracy_score))\n",
    "    models.append((tag, ovr_model))\n",
    "    \n",
    "end_time = time.time()\n",
    "execution_time = end_time - start_time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "796ccfcc-84ff-4c2c-94b6-557f5ac89385",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Execution time: 3566.5797970294952 seconds\n"
     ]
    }
   ],
   "source": [
    "print(f\"Execution time: {execution_time} seconds\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "323be979-0d80-406d-89f2-2b2a548b1158",
   "metadata": {},
   "source": [
    "**Note:** The executed models which are trained and tested on required data sets are stored at gs://msca-bdp-student-gcs/movie_synopsis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "2fd5f65f-1c70-47d8-b945-1e2074a8694b",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/27 20:43:22 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1391 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:43:30 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1395 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:43:37 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1401 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:43:42 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1405 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:43:49 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1411 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:43:52 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1415 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:43:57 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1421 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:44:01 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1425 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:44:07 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1431 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:44:10 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1435 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:44:15 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1441 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:44:18 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1445 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:44:23 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1451 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:44:26 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1455 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:44:31 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1461 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:44:35 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1465 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:44:40 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1471 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:44:43 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1475 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:44:48 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1481 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "23/11/27 20:44:51 WARN org.apache.spark.scheduler.TaskSetManager: Stage 1485 contains a task of very large size (6274 KiB). The maximum recommended task size is 1000 KiB.\n",
      "                                                                                \r"
     ]
    }
   ],
   "source": [
    "for tag, model in models:\n",
    "    model.write().overwrite().save(f'gs://msca-bdp-student-gcs/movie_synopsis/{tag}_model')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7e4dc507-63b7-43f4-877e-068aa899928d",
   "metadata": {},
   "source": [
    "***Displaying Test Scores***"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "18bb5b65-8b97-46d8-a376-f0aa232e684b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+----------------+--------+--------------+\n",
      "|           Model|F1 Score|Accuracy Score|\n",
      "+----------------+--------+--------------+\n",
      "|good versus evil|    0.91|          0.94|\n",
      "|     suspenseful|    0.88|          0.92|\n",
      "|          comedy|    0.82|          0.87|\n",
      "|     psychedelic|    0.82|          0.87|\n",
      "|         revenge|    0.76|          0.82|\n",
      "|            cult|    0.75|          0.81|\n",
      "|        romantic|    0.74|           0.8|\n",
      "|       flashback|    0.72|          0.78|\n",
      "|        violence|     0.7|          0.73|\n",
      "|          murder|    0.67|          0.69|\n",
      "+----------------+--------+--------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import pyspark.sql.functions as F\n",
    "\n",
    "from pyspark.sql.types import StructType, StructField, StringType, DoubleType\n",
    "\n",
    "schema = StructType([\n",
    "    StructField(\"Model\", StringType(), True),\n",
    "    StructField(\"F1 Score\", DoubleType(), True),\n",
    "    StructField(\"Accuracy Score\", DoubleType(), True)\n",
    "])\n",
    "\n",
    "score_df = spark.createDataFrame(results, schema=schema)\n",
    "\n",
    "score_df = score_df.withColumn(\"F1 Score\", F.round(F.col(\"F1 Score\"), 2))\n",
    "score_df = score_df.withColumn(\"Accuracy Score\", F.round(F.col(\"Accuracy Score\"), 2))\n",
    "\n",
    "score_df.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 217,
   "id": "4f81d5b3-2080-4844-9b5f-ace52294915b",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                                \r"
     ]
    }
   ],
   "source": [
    "from pyspark.ml.classification import OneVsRestModel\n",
    "\n",
    "# Load the models from GCS\n",
    "\n",
    "loaded_model = OneVsRestModel.load(f'gs://msca-bdp-student-gcs/movie_synopsis/murder_model')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7cfab047-b11e-4d46-892b-0afe29be9f40",
   "metadata": {},
   "source": [
    "## 3) Inference\n",
    "\n",
    "***Collect synopsis data for at least 10 recent movies (e.g., 2020 until today) from any public source. Using the NLP\n",
    "models created above ,identify which of the 10 popular tags apply to these movies. Note: if none of the top 10\n",
    "tags apply to your movies, explain why and investigate if the model can be further improved to generalize\n",
    "better.***"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "cdf24de3-b395-497c-9dfe-c70f10307b44",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                                \r"
     ]
    }
   ],
   "source": [
    "# load the models if classification was not re-run\n",
    "\n",
    "from pyspark.ml.classification import OneVsRestModel\n",
    "\n",
    "loaded_models = []\n",
    "for tag in top_tags:\n",
    "    \n",
    "    # Load the model from GCS\n",
    "    model = OneVsRestModel.load(f'gs://msca-bdp-student-gcs/movie_synopsis/{tag}_model')\n",
    "    loaded_models.append((tag, model))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0572105b-d676-4862-b798-69b459737dba",
   "metadata": {},
   "source": [
    "***Inference on New Movies Data***"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "6e517ea9-3e1f-4b7c-b123-849e520ff8a5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- plot_synopsis: string (nullable = true)\n",
      " |-- title: string (nullable = true)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "movies = spark.read.csv(\"gs://msca-bdp-student-gcs/movie_synopsis/movies_data.csv\",\n",
    "                    header=True, inferSchema= True, sep=\",\", quote='\"', escape='\"', multiLine=True)\n",
    "\n",
    "movies.printSchema()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "0e0450d0-9d95-4ebb-b492-8ed300f6477c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+--------------------+--------------------+\n",
      "|       plot_synopsis|               title|\n",
      "+--------------------+--------------------+\n",
      "|The story commenc...|         Oppenheimer|\n",
      "|The film begins w...|              Barbie|\n",
      "|During the First ...|The Hunger Games:...|\n",
      "|More than 30 year...|   Top gun: Maverick|\n",
      "|Evelyn Wang is go...|Everything Everyw...|\n",
      "|In 2002, Meilin \"...|         Turning Red|\n",
      "|The wishing star ...|Puss In Boots: Th...|\n",
      "|During the COVID-...|     The Glass Onion|\n",
      "|In 1992, Jonathan...|  Tick, Tick...BOOM!|\n",
      "|J. Jonah Jameson ...|Spiderman: No Way...|\n",
      "+--------------------+--------------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "movies.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "7a9914fc-c990-4837-a6a6-abd66f9c41d6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- plot_synopsis: string (nullable = true)\n",
      " |-- title: string (nullable = true)\n",
      " |-- token_plot: array (nullable = true)\n",
      " |    |-- element: string (containsNull = true)\n",
      " |-- processed_plot: array (nullable = true)\n",
      " |    |-- element: string (containsNull = true)\n",
      " |-- rawFeatures: vector (nullable = true)\n",
      " |-- tfidf_features: vector (nullable = true)\n",
      " |-- count_vectorized: vector (nullable = true)\n",
      " |-- features: vector (nullable = true)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# executing NLP data pipeline on the new movies dataset\n",
    "\n",
    "df_movies = fitted_pipeline.transform(movies)\n",
    "df_movies.printSchema()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "94a25b86-a1b1-48e9-a5a9-79b87d43cfdd",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- title: string (nullable = true)\n",
      " |-- features: vector (nullable = true)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df_movies = df_movies.select('title', 'features')\n",
    "df_movies.printSchema()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "9e7c96b6-0714-4924-8f97-fea7b371e1e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "new_results = []\n",
    "tag_results = {}\n",
    "\n",
    "for tag, model in models:  #change models to loaded_models if classification not re-executed\n",
    "\n",
    "    # Make predictions\n",
    "    predictions = model.transform(df_movies)\n",
    "\n",
    "    predictions = predictions.select(\"title\", \"prediction\")\n",
    "\n",
    "    # Rename the prediction column to the tag name\n",
    "    predictions = predictions.withColumnRenamed(\"prediction\", tag)\n",
    "\n",
    "    tag_results[tag] = predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "78a2b803-56fb-4bab-a5a2-60b7d0b50027",
   "metadata": {},
   "outputs": [],
   "source": [
    "# create dataframe of all predictions\n",
    "\n",
    "for tag, predictions_df in tag_results.items():\n",
    "    df_movies = df_movies.join(predictions_df, on=\"title\", how=\"left\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "f6e631d1-4638-40cc-b1cd-7b25b7100ffe",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/27 20:56:27 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.1 MiB\n",
      "23/11/27 20:56:28 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.1 MiB\n",
      "23/11/27 20:56:30 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.1 MiB\n",
      "23/11/27 20:56:32 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.1 MiB\n",
      "23/11/27 20:56:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.1 MiB\n",
      "23/11/27 20:56:36 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.1 MiB\n",
      "23/11/27 20:56:38 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.1 MiB\n",
      "23/11/27 20:56:40 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.1 MiB\n",
      "23/11/27 20:56:42 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.1 MiB\n",
      "23/11/27 20:56:44 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.1 MiB\n",
      "                                                                                \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+--------------------+----------------+-----------+------+-----------+-------+----+--------+---------+--------+------+\n",
      "|               title|good versus evil|suspenseful|comedy|psychedelic|revenge|cult|romantic|flashback|violence|murder|\n",
      "+--------------------+----------------+-----------+------+-----------+-------+----+--------+---------+--------+------+\n",
      "|         Oppenheimer|             0.0|        0.0|   0.0|        0.0|    0.0| 0.0|     0.0|      0.0|     0.0|   0.0|\n",
      "|              Barbie|             0.0|        0.0|   0.0|        0.0|    0.0| 0.0|     0.0|      0.0|     0.0|   0.0|\n",
      "|The Hunger Games:...|             0.0|        0.0|   0.0|        0.0|    0.0| 0.0|     0.0|      0.0|     0.0|   0.0|\n",
      "|   Top gun: Maverick|             0.0|        0.0|   0.0|        0.0|    0.0| 0.0|     0.0|      0.0|     0.0|   0.0|\n",
      "|Everything Everyw...|             0.0|        0.0|   0.0|        0.0|    0.0| 0.0|     0.0|      0.0|     0.0|   0.0|\n",
      "|         Turning Red|             0.0|        0.0|   0.0|        0.0|    0.0| 0.0|     0.0|      0.0|     0.0|   0.0|\n",
      "|Puss In Boots: Th...|             0.0|        0.0|   0.0|        0.0|    0.0| 0.0|     0.0|      0.0|     0.0|   0.0|\n",
      "|     The Glass Onion|             0.0|        0.0|   0.0|        0.0|    0.0| 0.0|     0.0|      0.0|     0.0|   0.0|\n",
      "|  Tick, Tick...BOOM!|             0.0|        0.0|   0.0|        0.0|    0.0| 0.0|     1.0|      0.0|     0.0|   0.0|\n",
      "|Spiderman: No Way...|             0.0|        0.0|   0.0|        0.0|    0.0| 0.0|     0.0|      0.0|     1.0|   0.0|\n",
      "+--------------------+----------------+-----------+------+-----------+-------+----+--------+---------+--------+------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "movies_tag_predictions = df_movies.drop('features')\n",
    "movies_tag_predictions.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8bcbb03c-182a-4130-9c99-a3bdc16a0b10",
   "metadata": {},
   "source": [
    "***Scores on Validation Set***"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "48d9f22f-be34-4ecc-9f08-3b2613aeb49a",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/27 21:11:56 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:12:38 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:13:25 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:14:13 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:14:59 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:15:41 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:16:44 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:17:18 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:18:02 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:18:40 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:19:18 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:20:01 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:21:00 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:21:34 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:22:07 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:22:36 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:23:11 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:23:43 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:24:20 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "23/11/27 21:24:49 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 67.2 MiB\n",
      "                                                                                \r"
     ]
    }
   ],
   "source": [
    "val_results = []\n",
    "\n",
    "for tag, model in models:  #change models to loaded_models if classification not re-executed\n",
    "    \n",
    "    eval_val = val.withColumnRenamed(tag, \"label\")\n",
    "    eval_val = eval_val.cache()\n",
    "    \n",
    "    # Make predictions on the validation data\n",
    "    predictions = model.transform(eval_val)\n",
    "\n",
    "    # Store F1 score in the dictionary\n",
    "    f1_score = evaluator_f1.evaluate(predictions)\n",
    "\n",
    "    # Store accuracy in the dictionary\n",
    "    accuracy_score = evaluator_accuracy.evaluate(predictions)\n",
    "\n",
    "    val_results.append((tag, f1_score, accuracy_score))\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "6afce924-81d8-4cb7-a442-381f305f3c3d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+----------------+--------+--------------+\n",
      "|           Model|F1 Score|Accuracy Score|\n",
      "+----------------+--------+--------------+\n",
      "|good versus evil|    0.92|          0.94|\n",
      "|     suspenseful|    0.88|          0.91|\n",
      "|          comedy|    0.82|          0.87|\n",
      "|     psychedelic|    0.81|          0.86|\n",
      "|         revenge|    0.76|          0.82|\n",
      "|            cult|    0.75|          0.81|\n",
      "|        romantic|    0.75|           0.8|\n",
      "|       flashback|    0.71|          0.78|\n",
      "|        violence|    0.73|          0.76|\n",
      "|          murder|    0.67|          0.69|\n",
      "+----------------+--------+--------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "val_score_df = spark.createDataFrame(val_results, schema=schema)\n",
    "\n",
    "val_score_df = val_score_df.withColumn(\"F1 Score\", F.round(F.col(\"F1 Score\"), 2))\n",
    "val_score_df = val_score_df.withColumn(\"Accuracy Score\", F.round(F.col(\"Accuracy Score\"), 2))\n",
    "\n",
    "val_score_df.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "30069e50-72e5-42f7-8388-116e240b7900",
   "metadata": {},
   "source": [
    "**Answer:** There are around 8 movies in the new dataset for which no tag was predicted by the models. The models did not perform well on this limited data set but when the model was evaluated on the validation data, it acheived similar F1 and accuracy scores as it did on the test data. It is interesting to note that 'murder' despite being the most common tag has the least F1 and accuracy scores on both test and validation data. \n",
    "\n",
    "I tried experimenting with different max iteration parameters (results not shown here) but the performance was similar. One of the reasons why none of the tags were applied to the new movies could be that since a portion of the training data contains tags that are not in the top ten, the model learns that representation too. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "e952c95e-4a46-4f78-86de-74eb98398f4e",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/28 02:03:49 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "                                                                                \r"
     ]
    },
    {
     "data": {
      "text/plain": [
       "9447"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "e9471ae1-2536-4c62-96ef-a4323fc0a6fd",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "23/11/28 02:07:16 WARN org.apache.spark.scheduler.DAGScheduler: Broadcasting large task binary with size 48.2 MiB\n",
      "                                                                                \r"
     ]
    },
    {
     "data": {
      "text/plain": [
       "1359"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_zeros = train.filter((F.col(\"good versus evil\") == 0) & (F.col(\"suspenseful\") == 0) & (F.col(\"comedy\") == 0) & (F.col(\"psychedelic\") == 0)\\\n",
    "            & (F.col(\"revenge\") == 0) & (F.col(\"cult\") == 0) & (F.col(\"romantic\") == 0) & (F.col(\"flashback\") == 0)& (F.col(\"violence\") == 0)\\\n",
    "            & (F.col(\"murder\") == 0)).count()\n",
    "all_zeros"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d79026e3-d7b7-4cf3-b61e-501381b3d70b",
   "metadata": {},
   "source": [
    "Another reason for predicting mostly no tags could be that a lot of movies in training might have only one tag associated, so it could lead to predicting 0 more number of times. Another interpretation of the zeroes can be that for some of the new movies, some tags might actually not be applicable to the movie in queston. For example, Tick Tick...Boom! is not a suspeseful, murder, cult or violent movie so having 0 in the predicted tag makes sense here and it does have a romantic angle that the main character struggles with so that tag though not accurate is not entirely wrong. On the other hand, Spiderman does have a violence component so it getting associated with that tag seems appropriate enough.\n",
    "\n",
    "The model could be improved for generalization by incorporating techniques that learn better representations of words in a corpus for NLP classification tasks such as Pointwise Mutual Information, creating word2vec embeddings, or using pre-trained embeddings like BERT, GPT etc. Fine-tuning the hyperparameters of the models by experimenting with different regularization parameters and increasing the maximum iterations it takes for the regression to converge could improve the generalization capabilities of the model. But the results will still not be as good as what can be achieved with advanced and more suitable NLP techniques."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a63505bb-2c1a-4cc4-bdcf-925ab4c2ef27",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "PySpark",
   "language": "python",
   "name": "pyspark"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}