1_notmnist.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "5hIbr52I7Z7U"
   },
   "source": [
    "Deep Learning\n",
    "=============\n",
    "\n",
    "Assignment 1\n",
    "------------\n",
    "\n",
    "The objective of this assignment is to learn about simple data curation practices, and familiarize you with some of the data we'll be reusing later.\n",
    "\n",
    "This notebook uses the [notMNIST](http://yaroslavvb.blogspot.com/2011/09/notmnist-dataset.html) dataset to be used with python experiments. This dataset is designed to look like the classic [MNIST](http://yann.lecun.com/exdb/mnist/) dataset, while looking a little more like real data: it's a harder task, and the data is a lot less 'clean' than MNIST."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     }
    },
    "colab_type": "code",
    "collapsed": false,
    "id": "apJbCsBHl-2A"
   },
   "outputs": [],
   "source": [
    "# These are all the modules we'll be using later. Make sure you can import them\n",
    "# before proceeding further.\n",
    "from __future__ import print_function\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import os\n",
    "import random\n",
    "import hashlib\n",
    "import json\n",
    "import sys\n",
    "import tarfile\n",
    "from IPython.display import display, Image\n",
    "from scipy import ndimage\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from six.moves.urllib.request import urlretrieve\n",
    "from six.moves import cPickle as pickle\n",
    "\n",
    "# Config the matlotlib backend as plotting inline in IPython\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "jNWGtZaXn-5j"
   },
   "source": [
    "First, we'll download the dataset to our local machine. The data consists of characters rendered in a variety of fonts on a 28x28 image. The labels are limited to 'A' through 'J' (10 classes). The training set has about 500k and the testset 19000 labelled examples. Given these sizes, it should be possible to train models quickly on any machine."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     },
     "output_extras": [
      {
       "item_id": 1
      }
     ]
    },
    "colab_type": "code",
    "collapsed": false,
    "executionInfo": {
     "elapsed": 186058,
     "status": "ok",
     "timestamp": 1444485672507,
     "user": {
      "color": "#1FA15D",
      "displayName": "Vincent Vanhoucke",
      "isAnonymous": false,
      "isMe": true,
      "permissionId": "05076109866853157986",
      "photoUrl": "//lh6.googleusercontent.com/-cCJa7dTDcgQ/AAAAAAAAAAI/AAAAAAAACgw/r2EZ_8oYer4/s50-c-k-no/photo.jpg",
      "sessionId": "2a0a5e044bb03b66",
      "userId": "102167687554210253930"
     },
     "user_tz": 420
    },
    "id": "EYRJ4ICW6-da",
    "outputId": "0d0f85df-155f-4a89-8e7e-ee32df36ec8d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found and verified notMNIST_large.tar.gz\n",
      "Found and verified notMNIST_small.tar.gz\n"
     ]
    }
   ],
   "source": [
    "url = 'http://commondatastorage.googleapis.com/books1000/'\n",
    "\n",
    "def maybe_download(filename, expected_bytes, force=False):\n",
    "  \"\"\"Download a file if not present, and make sure it's the right size.\"\"\"\n",
    "  if force or not os.path.exists(filename):\n",
    "    filename, _ = urlretrieve(url + filename, filename)\n",
    "  statinfo = os.stat(filename)\n",
    "  if statinfo.st_size == expected_bytes:\n",
    "    print('Found and verified', filename)\n",
    "  else:\n",
    "    raise Exception(\n",
    "      'Failed to verify ' + filename + '. Can you get to it with a browser?')\n",
    "  return filename\n",
    "\n",
    "train_filename = maybe_download('notMNIST_large.tar.gz', 247336696)\n",
    "test_filename = maybe_download('notMNIST_small.tar.gz', 8458043)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "cC3p0oEyF8QT"
   },
   "source": [
    "Extract the dataset from the compressed .tar.gz file.\n",
    "This should give you a set of directories, labelled A through J."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     },
     "output_extras": [
      {
       "item_id": 1
      }
     ]
    },
    "colab_type": "code",
    "collapsed": false,
    "executionInfo": {
     "elapsed": 186055,
     "status": "ok",
     "timestamp": 1444485672525,
     "user": {
      "color": "#1FA15D",
      "displayName": "Vincent Vanhoucke",
      "isAnonymous": false,
      "isMe": true,
      "permissionId": "05076109866853157986",
      "photoUrl": "//lh6.googleusercontent.com/-cCJa7dTDcgQ/AAAAAAAAAAI/AAAAAAAACgw/r2EZ_8oYer4/s50-c-k-no/photo.jpg",
      "sessionId": "2a0a5e044bb03b66",
      "userId": "102167687554210253930"
     },
     "user_tz": 420
    },
    "id": "H8CBE-WZ8nmj",
    "outputId": "ef6c790c-2513-4b09-962e-27c79390c762"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "notMNIST_large already present - Skipping extraction of notMNIST_large.tar.gz.\n",
      "['notMNIST_large/A', 'notMNIST_large/B', 'notMNIST_large/C', 'notMNIST_large/D', 'notMNIST_large/E', 'notMNIST_large/F', 'notMNIST_large/G', 'notMNIST_large/H', 'notMNIST_large/I', 'notMNIST_large/J']\n",
      "notMNIST_small already present - Skipping extraction of notMNIST_small.tar.gz.\n",
      "['notMNIST_small/A', 'notMNIST_small/B', 'notMNIST_small/C', 'notMNIST_small/D', 'notMNIST_small/E', 'notMNIST_small/F', 'notMNIST_small/G', 'notMNIST_small/H', 'notMNIST_small/I', 'notMNIST_small/J']\n"
     ]
    }
   ],
   "source": [
    "num_classes = 10\n",
    "np.random.seed(133)\n",
    "\n",
    "def maybe_extract(filename, force=False):\n",
    "  root = os.path.splitext(os.path.splitext(filename)[0])[0]  # remove .tar.gz\n",
    "  if os.path.isdir(root) and not force:\n",
    "    # You may override by setting force=True.\n",
    "    print('%s already present - Skipping extraction of %s.' % (root, filename))\n",
    "  else:\n",
    "    print('Extracting data for %s. This may take a while. Please wait.' % root)\n",
    "    tar = tarfile.open(filename)\n",
    "    sys.stdout.flush()\n",
    "    tar.extractall()\n",
    "    tar.close()\n",
    "  data_folders = [\n",
    "    os.path.join(root, d) for d in sorted(os.listdir(root))\n",
    "    if os.path.isdir(os.path.join(root, d))]\n",
    "  if len(data_folders) != num_classes:\n",
    "    raise Exception(\n",
    "      'Expected %d folders, one per class. Found %d instead.' % (\n",
    "        num_classes, len(data_folders)))\n",
    "  print(data_folders)\n",
    "  return data_folders\n",
    "  \n",
    "train_folders = maybe_extract(train_filename)\n",
    "test_folders = maybe_extract(test_filename)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "4riXK3IoHgx6"
   },
   "source": [
    "---\n",
    "Problem 1\n",
    "---------\n",
    "\n",
    "Let's take a peek at some of the data to make sure it looks sensible. Each exemplar should be an image of a character A through J rendered in a different font. Display a sample of the images that we just downloaded. Hint: you can use the package IPython.display.\n",
    "\n",
    "---"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAABwAAAAcCAAAAABXZoBIAAACGklEQVR4nG2SX0iTYRTGn3Pebyua\nzfjcSiWdWaSjBllZQ1h/qG0llDeiiRhEUhfRVcG8KqhL6yIKRkSSIWldjWF/bCxHFEo3FmPJbqKL\nQArZoiytr+99u9gf1+y5PD/e5zyc5wWKYqBuMHoUjJUi2AcifZ4xC61kTL6JCzbg+t7/PGU88AJW\n6hyAVrIoZyrr5bSVDZVoEmY5FHQoid9S8fzXFsVlUClvnLZUg+mVH2WQzA22d+puP0i9aLKY/0Km\ng2nDLevZFNl5b8G3aLsvhu4rP9ySKHYEVArJdK19Y980+T4AqJeuSpNKIKNjSgZSiO1kU1tMB4tj\nAJBiTxT7H+Pj91YFRPxQy5CVN/NpszWlYaIDipPatlwkzt382DME3zJRfGulyXh6QlEBknTUTWrN\nI9JQC4kAQONuXVIesuqeWtRrj++qrt3dcBKSf073KgZAAGjV0PkMgq0bdf42m44CcN7pWVIKAAS6\nLi3XBBAELp+ByJ9uuJGwzuGs0vUqp8MOIlofXcMEDcI8kPmA0z1ZoXLhbo8KfEmcG7QYACPcQmK4\npuBa86SCSawebwaxkL5fM+Qy5/IViLn4Wclq6erN7QTGPY8G77V8ABBXRJ1MjLbRRyzbs8k/nlCq\nUJOihbGQtDBmjM+gcENwJNKnofhdBR4eBvyvT0Hr9dmM+8+BfBEAFIXClh1t/bMCt254ABIoEaNx\n6CJDw19C+61FGl7e4gAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "A\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAABwAAAAcCAAAAABXZoBIAAACH0lEQVR4nE1RTUhUYRQ997vfTKWT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\n1Z/vng/PQR2B/4dc4cHo+n5kAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "B\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAABwAAAAcCAAAAABXZoBIAAAB80lEQVR4nF2SX2jOURjHP885v9d6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\n/WMu2uZDPe0ty/X72+Trh09/mksC/gLb+A4q9YNQoQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "C\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAABwAAAAcCAAAAABXZoBIAAACkklEQVR4nDWSXWjVdRjHP7/f/78N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\n5p87tyaTxniA/wHO7maGBqVghwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "D\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAABwAAAAcCAAAAABXZoBIAAABcUlEQVR4nGWSvWpVURCFv5m9FYKI\nYKUYbJKIiFGSmEDwIazzAIq1r2Mv4hMIdgG9hWgMVxsLERHEkIgKkqBnZlmcn+s9rm6fb681s2cO\ngHPpu1IjNXrigLFyLo2RjJct3ERjhjOpgLjVwxgS0r++r2Bx6jrefiplZv1yUMG0uIy1tx9PFqJF\n8ilA4Y5CkjLzxlzZCsY66QB2+Lna0JqiQrLRpZa330xzTsuF1RaKd6U2QMygVi620HgW8Y8Ro8TO\no+heMPlhSj+4f9ymV2BruLoNwKsTzy5WrA0wgKY8VTeSann+Kt3MQoD56/6MszVe13I/zOq5QRZA\n9uveTxOcfCCHtzzUH0kK7TFSbVhrU9L3S4m+rc544Upf/01YzDu5dlYGUJjW2TYzgcpNogKyo92m\nGdVks0/9/SD6nWR5/sITTu+3mx7pLhXq5aXeqaEbWe6SUFfPqINWh2784ycEfnsYx0xieuwCX+e/\nfx2xhwN/AbvIwkQNZB69AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "E\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAABwAAAAcCAAAAABXZoBIAAAAn0lEQVR4nMWRsQ0CMRAE985PgPiU\nIshogU4ogQa/AUKSr4GMAL1AyLckPL4zcghcZGu0e+u1EI2hnLXFAOAfsCvHKjeFDsqHq4M3i8pu\nktnMdHfsIzanvEz33AwURwBIyThe0/tiy8OQslNu/DpdQ7yt0cMH6KEvSyD4RfG+HlNGGE0W/imU\n/WlVxDoiA3xN5hapvVODsbEKZNWvfKmEJ9DoN4bHJFzOAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "F\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAABwAAAAcCAAAAABXZoBIAAAB8ElEQVR4nGWSP0iVURjGf+fPvRYo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\nA7jIwbHaQK/L1t6/XXj3C0ve3j/6HgLdtUioagAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "G\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAABwAAAAcCAAAAABXZoBIAAACaUlEQVR4nF2TT2hUdxDHPzPz3tvN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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "H\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAABwAAAAcCAAAAABXZoBIAAABzklEQVR4nG2SO4sTURSAv3PnJpug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\nTkSuQmCC\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "I\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAABwAAAAcCAAAAABXZoBIAAABUElEQVR4nG1RPy/DURQ97/WVEBIl\nFrGJ0EgsErOwCBZLJ2KoDyAmk/gIRotYu1iaNAZJRSId6IKFMBikJhURjfT37j0GET99707v5eT8\nufcAwTicU0jhbogZY68pbPMwxOCwplTh0wRsQFQsm8QQJ/ddATGDmTeq8n0aNmAS8wOJAS5ujAaq\nGLijUjgHE2bFIpXCWl8UrFDZ5gYyEWy8SfW8HYVF5yoOxZwaovycjcTpv6QXNvLhAeBQoKiwBBdx\nxBET0s9GolrkW1Tlmf39p4ZY6hEj2NfQEcg+UD3rw1HVFSXb3I7EQQZlivBxLLKHxdQLxfMA2Ygh\n9pj8FBkhDtbolZV0mt+nwxJ9Z5EmYwwAa1yJvqNI6wBYawxWqfLF9XSR49XNIcAgt/Wp4nncnVYt\nUFo31dP6B1XZ2On95zhUXJgccdDm69P1VbVpyT/wG28OroaCpplGAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "J\n"
     ]
    }
   ],
   "source": [
    "base_dir = os.getcwd() + \"/notMNIST_large/\"\n",
    "letters = [chr(ord('A') + i) for i in range(0,10) ]\n",
    "for letter in letters:\n",
    "    letter_dir = base_dir + letter\n",
    "    random_image = random.choice(os.listdir(letter_dir))\n",
    "    display(Image(filename=letter_dir+ '/' + random_image))\n",
    "    print(letter)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "PBdkjESPK8tw"
   },
   "source": [
    "Now let's load the data in a more manageable format. Since, depending on your computer setup you might not be able to fit it all in memory, we'll load each class into a separate dataset, store them on disk and curate them independently. Later we'll merge them into a single dataset of manageable size.\n",
    "\n",
    "We'll convert the entire dataset into a 3D array (image index, x, y) of floating point values, normalized to have approximately zero mean and standard deviation ~0.5 to make training easier down the road. \n",
    "\n",
    "A few images might not be readable, we'll just skip them."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     },
     "output_extras": [
      {
       "item_id": 30
      }
     ]
    },
    "colab_type": "code",
    "collapsed": false,
    "executionInfo": {
     "elapsed": 399874,
     "status": "ok",
     "timestamp": 1444485886378,
     "user": {
      "color": "#1FA15D",
      "displayName": "Vincent Vanhoucke",
      "isAnonymous": false,
      "isMe": true,
      "permissionId": "05076109866853157986",
      "photoUrl": "//lh6.googleusercontent.com/-cCJa7dTDcgQ/AAAAAAAAAAI/AAAAAAAACgw/r2EZ_8oYer4/s50-c-k-no/photo.jpg",
      "sessionId": "2a0a5e044bb03b66",
      "userId": "102167687554210253930"
     },
     "user_tz": 420
    },
    "id": "h7q0XhG3MJdf",
    "outputId": "92c391bb-86ff-431d-9ada-315568a19e59"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Pickling notMNIST_large/A.pickle.\n",
      "notMNIST_large/A\n",
      "Could not read: notMNIST_large/A/RnJlaWdodERpc3BCb29rSXRhbGljLnR0Zg==.png : cannot identify image file - it's ok, skipping.\n",
      "Could not read: notMNIST_large/A/Um9tYW5hIEJvbGQucGZi.png : cannot identify image file - it's ok, skipping.\n",
      "Could not read: notMNIST_large/A/SG90IE11c3RhcmQgQlROIFBvc3Rlci50dGY=.png : cannot identify image file - it's ok, skipping.\n",
      "Full dataset tensor: (52912, 28, 28)\n",
      "Mean: -0.128478\n",
      "Standard deviation: 0.425835\n",
      "Pickling notMNIST_large/B.pickle.\n",
      "notMNIST_large/B\n",
      "Could not read: notMNIST_large/B/TmlraXNFRi1TZW1pQm9sZEl0YWxpYy5vdGY=.png : cannot identify image file - it's ok, skipping.\n",
      "Full dataset tensor: (52912, 28, 28)\n",
      "Mean: -0.0075604\n",
      "Standard deviation: 0.417294\n",
      "Pickling notMNIST_large/C.pickle.\n",
      "notMNIST_large/C\n",
      "Full dataset tensor: (52912, 28, 28)\n",
      "Mean: -0.142319\n",
      "Standard deviation: 0.421427\n",
      "Pickling notMNIST_large/D.pickle.\n",
      "notMNIST_large/D\n",
      "Could not read: notMNIST_large/D/VHJhbnNpdCBCb2xkLnR0Zg==.png : cannot identify image file - it's ok, skipping.\n",
      "Full dataset tensor: (52912, 28, 28)\n",
      "Mean: -0.0574537\n",
      "Standard deviation: 0.434084\n",
      "Pickling notMNIST_large/E.pickle.\n",
      "notMNIST_large/E\n",
      "Full dataset tensor: (52912, 28, 28)\n",
      "Mean: -0.0701407\n",
      "Standard deviation: 0.428825\n",
      "Pickling notMNIST_large/F.pickle.\n",
      "notMNIST_large/F\n",
      "Full dataset tensor: (52912, 28, 28)\n",
      "Mean: -0.125928\n",
      "Standard deviation: 0.429618\n",
      "Pickling notMNIST_large/G.pickle.\n",
      "notMNIST_large/G\n",
      "Full dataset tensor: (52912, 28, 28)\n",
      "Mean: -0.0947753\n",
      "Standard deviation: 0.421783\n",
      "Pickling notMNIST_large/H.pickle.\n",
      "notMNIST_large/H\n",
      "Full dataset tensor: (52912, 28, 28)\n",
      "Mean: -0.0687492\n",
      "Standard deviation: 0.430445\n",
      "Pickling notMNIST_large/I.pickle.\n",
      "notMNIST_large/I\n",
      "Full dataset tensor: (52912, 28, 28)\n",
      "Mean: 0.0307373\n",
      "Standard deviation: 0.449686\n",
      "Pickling notMNIST_large/J.pickle.\n",
      "notMNIST_large/J\n",
      "Full dataset tensor: (52911, 28, 28)\n",
      "Mean: -0.153468\n",
      "Standard deviation: 0.397116\n",
      "Pickling notMNIST_small/A.pickle.\n",
      "notMNIST_small/A\n",
      "Could not read: notMNIST_small/A/RGVtb2NyYXRpY2FCb2xkT2xkc3R5bGUgQm9sZC50dGY=.png : cannot identify image file - it's ok, skipping.\n",
      "Full dataset tensor: (1873, 28, 28)\n",
      "Mean: -0.132516\n",
      "Standard deviation: 0.445818\n",
      "Pickling notMNIST_small/B.pickle.\n",
      "notMNIST_small/B\n",
      "Full dataset tensor: (1873, 28, 28)\n",
      "Mean: 0.00535639\n",
      "Standard deviation: 0.457048\n",
      "Pickling notMNIST_small/C.pickle.\n",
      "notMNIST_small/C\n",
      "Full dataset tensor: (1873, 28, 28)\n",
      "Mean: -0.141488\n",
      "Standard deviation: 0.441056\n",
      "Pickling notMNIST_small/D.pickle.\n",
      "notMNIST_small/D\n",
      "Full dataset tensor: (1873, 28, 28)\n",
      "Mean: -0.0492095\n",
      "Standard deviation: 0.46049\n",
      "Pickling notMNIST_small/E.pickle.\n",
      "notMNIST_small/E\n",
      "Full dataset tensor: (1873, 28, 28)\n",
      "Mean: -0.0599003\n",
      "Standard deviation: 0.456123\n",
      "Pickling notMNIST_small/F.pickle.\n",
      "notMNIST_small/F\n",
      "Could not read: notMNIST_small/F/Q3Jvc3NvdmVyIEJvbGRPYmxpcXVlLnR0Zg==.png : cannot identify image file - it's ok, skipping.\n",
      "Full dataset tensor: (1873, 28, 28)\n",
      "Mean: -0.118168\n",
      "Standard deviation: 0.451113\n",
      "Pickling notMNIST_small/G.pickle.\n",
      "notMNIST_small/G\n",
      "Full dataset tensor: (1872, 28, 28)\n",
      "Mean: -0.0925151\n",
      "Standard deviation: 0.44846\n",
      "Pickling notMNIST_small/H.pickle.\n",
      "notMNIST_small/H\n",
      "Full dataset tensor: (1872, 28, 28)\n",
      "Mean: -0.0586728\n",
      "Standard deviation: 0.457382\n",
      "Pickling notMNIST_small/I.pickle.\n",
      "notMNIST_small/I\n",
      "Full dataset tensor: (1872, 28, 28)\n",
      "Mean: 0.0526455\n",
      "Standard deviation: 0.472764\n",
      "Pickling notMNIST_small/J.pickle.\n",
      "notMNIST_small/J\n",
      "Full dataset tensor: (1872, 28, 28)\n",
      "Mean: -0.151675\n",
      "Standard deviation: 0.449515\n"
     ]
    }
   ],
   "source": [
    "image_size = 28  # Pixel width and height.\n",
    "pixel_depth = 255.0  # Number of levels per pixel.\n",
    "\n",
    "def load_letter(folder, min_num_images):\n",
    "  \"\"\"Load the data for a single letter label.\"\"\"\n",
    "  image_files = os.listdir(folder)\n",
    "  dataset = np.ndarray(shape=(len(image_files), image_size, image_size),\n",
    "                         dtype=np.float32)\n",
    "  print(folder)\n",
    "  for image_index, image in enumerate(image_files):\n",
    "    image_file = os.path.join(folder, image)\n",
    "    try:\n",
    "      image_data = (ndimage.imread(image_file).astype(float) - \n",
    "                    pixel_depth / 2) / pixel_depth\n",
    "      if image_data.shape != (image_size, image_size):\n",
    "        raise Exception('Unexpected image shape: %s' % str(image_data.shape))\n",
    "      dataset[image_index, :, :] = image_data\n",
    "    except IOError as e:\n",
    "      print('Could not read:', image_file, ':', e, '- it\\'s ok, skipping.')\n",
    "    \n",
    "  num_images = image_index + 1\n",
    "  dataset = dataset[0:num_images, :, :]\n",
    "  if num_images < min_num_images:\n",
    "    raise Exception('Many fewer images than expected: %d < %d' %\n",
    "                    (num_images, min_num_images))\n",
    "    \n",
    "  print('Full dataset tensor:', dataset.shape)\n",
    "  print('Mean:', np.mean(dataset))\n",
    "  print('Standard deviation:', np.std(dataset))\n",
    "  return dataset\n",
    "        \n",
    "def maybe_pickle(data_folders, min_num_images_per_class, force=False):\n",
    "  dataset_names = []\n",
    "  for folder in data_folders:\n",
    "    set_filename = folder + '.pickle'\n",
    "    dataset_names.append(set_filename)\n",
    "    if os.path.exists(set_filename) and not force:\n",
    "      # You may override by setting force=True.\n",
    "      print('%s already present - Skipping pickling.' % set_filename)\n",
    "    else:\n",
    "      print('Pickling %s.' % set_filename)\n",
    "      dataset = load_letter(folder, min_num_images_per_class)\n",
    "      try:\n",
    "        with open(set_filename, 'wb') as f:\n",
    "          pickle.dump(dataset, f, pickle.HIGHEST_PROTOCOL)\n",
    "      except Exception as e:\n",
    "        print('Unable to save data to', set_filename, ':', e)\n",
    "  \n",
    "  return dataset_names\n",
    "\n",
    "train_datasets = maybe_pickle(train_folders, 45000)\n",
    "test_datasets = maybe_pickle(test_folders, 1800)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "vUdbskYE2d87"
   },
   "source": [
    "---\n",
    "Problem 2\n",
    "---------\n",
    "\n",
    "Let's verify that the data still looks good. Displaying a sample of the labels and images from the ndarray. Hint: you can use matplotlib.pyplot.\n",
    "\n",
    "---"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f53b85b8690>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP4AAAD8CAYAAABXXhlaAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f53c2f53f90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "A_list = pickle.load(open(\"notMNIST_large/A.pickle\", \"rb\"))\n",
    "random_letter = random.choice(A_list)\n",
    "%matplotlib inline\n",
    "plt.imshow(random_letter)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "cYznx5jUwzoO"
   },
   "source": [
    "---\n",
    "Problem 3\n",
    "---------\n",
    "Another check: we expect the data to be balanced across classes. Verify that.\n",
    "\n",
    "---"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "A train data count : 52912\n",
      "A test data count : 1873\n",
      "B train data count : 52912\n",
      "B test data count : 1873\n",
      "C train data count : 52912\n",
      "C test data count : 1873\n",
      "D train data count : 52912\n",
      "D test data count : 1873\n",
      "E train data count : 52912\n",
      "E test data count : 1873\n",
      "F train data count : 52912\n",
      "F test data count : 1873\n",
      "G train data count : 52912\n",
      "G test data count : 1872\n",
      "H train data count : 52912\n",
      "H test data count : 1872\n",
      "I train data count : 52912\n",
      "I test data count : 1872\n",
      "J train data count : 52911\n",
      "J test data count : 1872\n"
     ]
    }
   ],
   "source": [
    "letters = [chr(ord('A') + i) for i in range(0,10) ]\n",
    "for letter in letters:\n",
    "    letter_train_data = pickle.load(open('notMNIST_large/' + letter + '.pickle', \"rb\"))\n",
    "    print(letter + \" train data count : \" + str(len(letter_train_data)) )\n",
    "    letter_test_data = pickle.load(open('notMNIST_small/' + letter + '.pickle', \"rb\"))\n",
    "    print(letter + \" test data count : \" + str(len(letter_test_data)) )\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "LA7M7K22ynCt"
   },
   "source": [
    "Merge and prune the training data as needed. Depending on your computer setup, you might not be able to fit it all in memory, and you can tune `train_size` as needed. The labels will be stored into a separate array of integers 0 through 9.\n",
    "\n",
    "Also create a validation dataset for hyperparameter tuning."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     },
     "output_extras": [
      {
       "item_id": 1
      }
     ]
    },
    "colab_type": "code",
    "collapsed": false,
    "executionInfo": {
     "elapsed": 411281,
     "status": "ok",
     "timestamp": 1444485897869,
     "user": {
      "color": "#1FA15D",
      "displayName": "Vincent Vanhoucke",
      "isAnonymous": false,
      "isMe": true,
      "permissionId": "05076109866853157986",
      "photoUrl": "//lh6.googleusercontent.com/-cCJa7dTDcgQ/AAAAAAAAAAI/AAAAAAAACgw/r2EZ_8oYer4/s50-c-k-no/photo.jpg",
      "sessionId": "2a0a5e044bb03b66",
      "userId": "102167687554210253930"
     },
     "user_tz": 420
    },
    "id": "s3mWgZLpyuzq",
    "outputId": "8af66da6-902d-4719-bedc-7c9fb7ae7948"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training: (200000, 28, 28) (200000,)\n",
      "Validation: (10000, 28, 28) (10000,)\n",
      "Testing: (10000, 28, 28) (10000,)\n"
     ]
    }
   ],
   "source": [
    "def make_arrays(nb_rows, img_size):\n",
    "  if nb_rows:\n",
    "    dataset = np.ndarray((nb_rows, img_size, img_size), dtype=np.float32)\n",
    "    labels = np.ndarray(nb_rows, dtype=np.int32)\n",
    "  else:\n",
    "    dataset, labels = None, None\n",
    "  return dataset, labels\n",
    "\n",
    "def merge_datasets(pickle_files, train_size, valid_size=0):\n",
    "  num_classes = len(pickle_files)\n",
    "  valid_dataset, valid_labels = make_arrays(valid_size, image_size)\n",
    "  train_dataset, train_labels = make_arrays(train_size, image_size)\n",
    "  vsize_per_class = valid_size // num_classes\n",
    "  tsize_per_class = train_size // num_classes\n",
    "    \n",
    "  start_v, start_t = 0, 0\n",
    "  end_v, end_t = vsize_per_class, tsize_per_class\n",
    "  end_l = vsize_per_class+tsize_per_class\n",
    "  for label, pickle_file in enumerate(pickle_files):       \n",
    "    try:\n",
    "      with open(pickle_file, 'rb') as f:\n",
    "        letter_set = pickle.load(f)\n",
    "        # let's shuffle the letters to have random validation and training set\n",
    "        np.random.shuffle(letter_set)\n",
    "        if valid_dataset is not None:\n",
    "          valid_letter = letter_set[:vsize_per_class, :, :]\n",
    "          valid_dataset[start_v:end_v, :, :] = valid_letter\n",
    "          valid_labels[start_v:end_v] = label\n",
    "          start_v += vsize_per_class\n",
    "          end_v += vsize_per_class\n",
    "                    \n",
    "        train_letter = letter_set[vsize_per_class:end_l, :, :]\n",
    "        train_dataset[start_t:end_t, :, :] = train_letter\n",
    "        train_labels[start_t:end_t] = label\n",
    "        start_t += tsize_per_class\n",
    "        end_t += tsize_per_class\n",
    "    except Exception as e:\n",
    "      print('Unable to process data from', pickle_file, ':', e)\n",
    "      raise\n",
    "    \n",
    "  return valid_dataset, valid_labels, train_dataset, train_labels\n",
    "            \n",
    "            \n",
    "train_size = 200000\n",
    "valid_size = 10000\n",
    "test_size = 10000\n",
    "\n",
    "valid_dataset, valid_labels, train_dataset, train_labels = merge_datasets(\n",
    "  train_datasets, train_size, valid_size)\n",
    "_, _, test_dataset, test_labels = merge_datasets(test_datasets, test_size)\n",
    "\n",
    "print('Training:', train_dataset.shape, train_labels.shape)\n",
    "print('Validation:', valid_dataset.shape, valid_labels.shape)\n",
    "print('Testing:', test_dataset.shape, test_labels.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "GPTCnjIcyuKN"
   },
   "source": [
    "Next, we'll randomize the data. It's important to have the labels well shuffled for the training and test distributions to match."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     }
    },
    "colab_type": "code",
    "collapsed": false,
    "id": "6WZ2l2tN2zOL"
   },
   "outputs": [],
   "source": [
    "def randomize(dataset, labels):\n",
    "  permutation = np.random.permutation(labels.shape[0])\n",
    "  shuffled_dataset = dataset[permutation,:,:]\n",
    "  shuffled_labels = labels[permutation]\n",
    "  return shuffled_dataset, shuffled_labels\n",
    "train_dataset, train_labels = randomize(train_dataset, train_labels)\n",
    "test_dataset, test_labels = randomize(test_dataset, test_labels)\n",
    "valid_dataset, valid_labels = randomize(valid_dataset, valid_labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "puDUTe6t6USl"
   },
   "source": [
    "---\n",
    "Problem 4\n",
    "---------\n",
    "Convince yourself that the data is still good after shuffling!\n",
    "\n",
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "tIQJaJuwg5Hw"
   },
   "source": [
    "Finally, let's save the data for later reuse:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     }
    },
    "colab_type": "code",
    "collapsed": true,
    "id": "QiR_rETzem6C"
   },
   "outputs": [],
   "source": [
    "pickle_file = 'notMNIST.pickle'\n",
    "\n",
    "try:\n",
    "  f = open(pickle_file, 'wb')\n",
    "  save = {\n",
    "    'train_dataset': train_dataset,\n",
    "    'train_labels': train_labels,\n",
    "    'valid_dataset': valid_dataset,\n",
    "    'valid_labels': valid_labels,\n",
    "    'test_dataset': test_dataset,\n",
    "    'test_labels': test_labels,\n",
    "    }\n",
    "  pickle.dump(save, f, pickle.HIGHEST_PROTOCOL)\n",
    "  f.close()\n",
    "except Exception as e:\n",
    "  print('Unable to save data to', pickle_file, ':', e)\n",
    "  raise"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     },
     "output_extras": [
      {
       "item_id": 1
      }
     ]
    },
    "colab_type": "code",
    "collapsed": false,
    "executionInfo": {
     "elapsed": 413065,
     "status": "ok",
     "timestamp": 1444485899688,
     "user": {
      "color": "#1FA15D",
      "displayName": "Vincent Vanhoucke",
      "isAnonymous": false,
      "isMe": true,
      "permissionId": "05076109866853157986",
      "photoUrl": "//lh6.googleusercontent.com/-cCJa7dTDcgQ/AAAAAAAAAAI/AAAAAAAACgw/r2EZ_8oYer4/s50-c-k-no/photo.jpg",
      "sessionId": "2a0a5e044bb03b66",
      "userId": "102167687554210253930"
     },
     "user_tz": 420
    },
    "id": "hQbLjrW_iT39",
    "outputId": "b440efc6-5ee1-4cbc-d02d-93db44ebd956"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Compressed pickle size: 690800441\n"
     ]
    }
   ],
   "source": [
    "statinfo = os.stat(pickle_file)\n",
    "print('Compressed pickle size:', statinfo.st_size)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "gE_cRAQB33lk"
   },
   "source": [
    "---\n",
    "Problem 5\n",
    "---------\n",
    "\n",
    "By construction, this dataset might contain a lot of overlapping samples, including training data that's also contained in the validation and test set! Overlap between training and test can skew the results if you expect to use your model in an environment where there is never an overlap, but are actually ok if you expect to see training samples recur when you use it.\n",
    "Measure how much overlap there is between training, validation and test samples.\n",
    "\n",
    "Optional questions:\n",
    "- What about near duplicates between datasets? (images that are almost identical)\n",
    "- Create a sanitized validation and test set, and compare your accuracy on those in subsequent assignments.\n",
    "---"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "188\n",
      "3435\n",
      "3428\n"
     ]
    }
   ],
   "source": [
    "all_data = pickle.load(open('notMNIST.pickle', 'rb'))\n",
    "\n",
    "def count_duplicates(dataset1, dataset2):\n",
    "    hashes = [hashlib.sha1(x).hexdigest() for x in dataset1]\n",
    "    dup_indices = []\n",
    "    for i in range(0, len(dataset2)):\n",
    "        if hashlib.sha1(dataset2[i]).hexdigest() in hashes:\n",
    "            dup_indices.append(i)\n",
    "    return len(dup_indices)\n",
    "\n",
    "\n",
    "print(count_duplicates(all_data['test_dataset'], all_data['valid_dataset']))\n",
    "print(count_duplicates(all_data['valid_dataset'], all_data['train_dataset']))\n",
    "print(count_duplicates(all_data['test_dataset'], all_data['train_dataset']))\n",
    "\n",
    "\n",
    "    \n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "L8oww1s4JMQx"
   },
   "source": [
    "---\n",
    "Problem 6\n",
    "---------\n",
    "\n",
    "Let's get an idea of what an off-the-shelf classifier can give you on this data. It's always good to check that there is something to learn, and that it's a problem that is not so trivial that a canned solution solves it.\n",
    "\n",
    "Train a simple model on this data using 50, 100, 1000 and 5000 training samples. Hint: you can use the LogisticRegression model from sklearn.linear_model.\n",
    "\n",
    "Optional question: train an off-the-shelf model on all the data!\n",
    "\n",
    "---"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "train_dataset = all_data['train_dataset']\n",
    "train_labels = all_data['train_labels']\n",
    "test_dataset = all_data['test_dataset']\n",
    "test_labels = all_data['test_labels']\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100 trainsamples score: 0.7657\n",
      "1000 trainsamples score: 0.8349\n",
      "5000 trainsamples score: 0.8489\n",
      "10000 trainsamples score: 0.8644\n",
      "all trainsamples score: 0.8935\n"
     ]
    }
   ],
   "source": [
    "\n",
    "def get_score(train_dataset, train_labels, test_dataset, test_labels):\n",
    "    model = LogisticRegression()\n",
    "    train_flatten_dataset = np.array([x.flatten() for x in train_dataset])\n",
    "    test_flatten_dataset = np.array([x.flatten() for x in test_dataset])\n",
    "    model.fit(train_flatten_dataset, train_labels)\n",
    "\n",
    "    return model.score([x.flatten() for x in test_dataset], test_labels)\n",
    "\n",
    "print(\"100 trainsamples score: \" + str(get_score(train_dataset[:100], train_labels[:100], test_dataset, test_labels)))\n",
    "print(\"1000 trainsamples score: \" + str(get_score(train_dataset[:1000], train_labels[:1000], test_dataset, test_labels)))\n",
    "print(\"5000 trainsamples score: \" + str(get_score(train_dataset[:5000], train_labels[:5000], test_dataset, test_labels)))\n",
    "print(\"10000 trainsamples score: \" + str(get_score(train_dataset[:10000], train_labels[:10000], test_dataset, test_labels)))\n",
    "print(\"all trainsamples score: \" + str(get_score(train_dataset, train_labels, test_dataset, test_labels)))\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "default_view": {},
   "name": "1_notmnist.ipynb",
   "provenance": [],
   "version": "0.3.2",
   "views": {}
  },
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}