StackOverflow-Data-Analysis

Analysing certain aspects of the stack overflow data

Methodology

1) Hive & Pig

Query 1: Most Viewed Users

Writing query for top 100 most viewed users in the user dataset, column referred in this dataset was views.

Query 2: Most Valuable Users

Writing query for top 100 most valuable users in the user dataset, column referred in this was reputation which signifies the repute of a user.

Query 3: Accepted Answer Percentage

Writing query for percentage of accepted answer by joining post answers and post questions and then finding the ration between accepted answers and total answers.

Query 4: Marking Spammers

Writing query for marking spammers by matching vote id from votes dataset id = 12 (as described by Stack Overflow community a spam) with user id in comments dataset and grouping into one table.

2) Tag Prediction

Step 1: Importing libraris relating to PySpark

Step 2: Reading the .csv file into PySpark Data Frame

Step 3: Splitting the data into train and test with a ration of 75:25

Step 4: Removing HTML tags from posts using BsTextExtractor

Step 5: Creating list of stopwords

Step 6: Defining Variables for pipeline

Step 7: Building pipeline

Step 8: Fitting the model

Step 9: Model predicting

Step 10: Model Evaluation