notebooks/logistic_regression_customdata.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "\"\"\"\n",
    " Logistic Regression with Custom Data\n",
    " @Sungjoon Choi (sungjoon.choi@cpslab.snu.ac.kr)\n",
    "\"\"\"\n",
    "import os\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "69 train images loaded\n",
      "18 test images loaded\n",
      "900 dimensional input\n",
      "2 classes\n"
     ]
    }
   ],
   "source": [
    "# Load them!\n",
    "cwd = os.getcwd()\n",
    "loadpath = cwd + \"/data/trainingset.npz\"\n",
    "l = np.load(loadpath)\n",
    "\n",
    "# See what's in here\n",
    "l.files\n",
    "\n",
    "# Parse data\n",
    "trainimg = l['trainimg']\n",
    "trainlabel = l['trainlabel']\n",
    "testimg = l['testimg']\n",
    "testlabel = l['testlabel']\n",
    "ntrain = trainimg.shape[0]\n",
    "nclass = trainlabel.shape[1]\n",
    "dim    = trainimg.shape[1]\n",
    "ntest  = testimg.shape[0]\n",
    "\n",
    "print (\"%d train images loaded\" % (ntrain))\n",
    "print (\"%d test images loaded\" % (ntest))\n",
    "print (\"%d dimensional input\" % (dim))\n",
    "print (\"%d classes\" % (nclass))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Network Ready\n"
     ]
    }
   ],
   "source": [
    "# Parameters of Logistic Regression\n",
    "learning_rate   = 0.001\n",
    "training_epochs = 1000\n",
    "batch_size      = 10\n",
    "display_step    = 10\n",
    "\n",
    "# Create Graph for Logistic Regression\n",
    "x = tf.placeholder(\"float\", [None, dim]) \n",
    "y = tf.placeholder(\"float\", [None, nclass])  # None is for infinite or unspecified length\n",
    "W = tf.Variable(tf.zeros([dim, nclass]))\n",
    "b = tf.Variable(tf.zeros([nclass]))\n",
    "\n",
    "# Activation, Cost, and Optimizing functions\n",
    "_pred = tf.nn.softmax(tf.matmul(x, W) + b) # Softmax\n",
    "cost = tf.reduce_mean(-tf.reduce_sum(y*tf.log(_pred), reduction_indices=1)) # Cross entropy\n",
    "# cost = tf.reduce_mean(tf.pow(activation-y, 2))\n",
    "optm = tf.train.GradientDescentOptimizer(learning_rate).minimize(cost) # Gradient Descent\n",
    "_corr = tf.equal(tf.argmax(_pred, 1), tf.argmax(y, 1))    \n",
    "accr = tf.reduce_mean(tf.cast(_corr, tf.float32))\n",
    "init = tf.initialize_all_variables()\n",
    "\n",
    "print (\"Network Ready\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 000/1000 cost: 0.677713056\n",
      " Training accuracy: 0.600\n",
      " Test accuracy: 0.444\n",
      "Epoch: 010/1000 cost: 0.647019178\n",
      " Training accuracy: 0.400\n",
      " Test accuracy: 0.389\n",
      "Epoch: 020/1000 cost: 0.620272974\n",
      " Training accuracy: 0.500\n",
      " Test accuracy: 0.444\n",
      "Epoch: 030/1000 cost: 0.596143623\n",
      " Training accuracy: 0.600\n",
      " Test accuracy: 0.389\n",
      "Epoch: 040/1000 cost: 0.610756665\n",
      " Training accuracy: 0.300\n",
      " Test accuracy: 0.389\n",
      "Epoch: 050/1000 cost: 0.597485046\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.500\n",
      "Epoch: 060/1000 cost: 0.563908676\n",
      " Training accuracy: 0.800\n",
      " Test accuracy: 0.444\n",
      "Epoch: 070/1000 cost: 0.561829552\n",
      " Training accuracy: 0.700\n",
      " Test accuracy: 0.500\n",
      "Epoch: 080/1000 cost: 0.569923252\n",
      " Training accuracy: 0.800\n",
      " Test accuracy: 0.500\n",
      "Epoch: 090/1000 cost: 0.588533024\n",
      " Training accuracy: 0.800\n",
      " Test accuracy: 0.444\n",
      "Epoch: 100/1000 cost: 0.552061776\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.556\n",
      "Epoch: 110/1000 cost: 0.537217483\n",
      " Training accuracy: 0.800\n",
      " Test accuracy: 0.556\n",
      "Epoch: 120/1000 cost: 0.546461110\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.556\n",
      "Epoch: 130/1000 cost: 0.505996898\n",
      " Training accuracy: 0.700\n",
      " Test accuracy: 0.556\n",
      "Epoch: 140/1000 cost: 0.495907933\n",
      " Training accuracy: 0.700\n",
      " Test accuracy: 0.500\n",
      "Epoch: 150/1000 cost: 0.534719969\n",
      " Training accuracy: 0.700\n",
      " Test accuracy: 0.556\n",
      "Epoch: 160/1000 cost: 0.513395478\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.556\n",
      "Epoch: 170/1000 cost: 0.478472695\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.556\n",
      "Epoch: 180/1000 cost: 0.517475565\n",
      " Training accuracy: 0.800\n",
      " Test accuracy: 0.556\n",
      "Epoch: 190/1000 cost: 0.485752508\n",
      " Training accuracy: 0.800\n",
      " Test accuracy: 0.444\n",
      "Epoch: 200/1000 cost: 0.515886277\n",
      " Training accuracy: 0.800\n",
      " Test accuracy: 0.611\n",
      "Epoch: 210/1000 cost: 0.499649867\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.556\n",
      "Epoch: 220/1000 cost: 0.472772673\n",
      " Training accuracy: 0.700\n",
      " Test accuracy: 0.611\n",
      "Epoch: 230/1000 cost: 0.451913580\n",
      " Training accuracy: 0.800\n",
      " Test accuracy: 0.556\n",
      "Epoch: 240/1000 cost: 0.479279583\n",
      " Training accuracy: 0.700\n",
      " Test accuracy: 0.556\n",
      "Epoch: 250/1000 cost: 0.453433091\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.611\n",
      "Epoch: 260/1000 cost: 0.455181931\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.611\n",
      "Epoch: 270/1000 cost: 0.430772195\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.556\n",
      "Epoch: 280/1000 cost: 0.451103106\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.556\n",
      "Epoch: 290/1000 cost: 0.454624414\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.611\n",
      "Epoch: 300/1000 cost: 0.455677142\n",
      " Training accuracy: 0.800\n",
      " Test accuracy: 0.611\n",
      "Epoch: 310/1000 cost: 0.482660721\n",
      " Training accuracy: 0.600\n",
      " Test accuracy: 0.556\n",
      "Epoch: 320/1000 cost: 0.482767885\n",
      " Training accuracy: 0.800\n",
      " Test accuracy: 0.611\n",
      "Epoch: 330/1000 cost: 0.473389422\n",
      " Training accuracy: 0.700\n",
      " Test accuracy: 0.611\n",
      "Epoch: 340/1000 cost: 0.471625254\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.667\n",
      "Epoch: 350/1000 cost: 0.387609541\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.611\n",
      "Epoch: 360/1000 cost: 0.395270055\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.556\n",
      "Epoch: 370/1000 cost: 0.377154971\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.667\n",
      "Epoch: 380/1000 cost: 0.453811030\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.611\n",
      "Epoch: 390/1000 cost: 0.377268990\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.611\n",
      "Epoch: 400/1000 cost: 0.451625079\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.556\n",
      "Epoch: 410/1000 cost: 0.406180809\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.556\n",
      "Epoch: 420/1000 cost: 0.397555699\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.556\n",
      "Epoch: 430/1000 cost: 0.387203733\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.556\n",
      "Epoch: 440/1000 cost: 0.435563892\n",
      " Training accuracy: 0.800\n",
      " Test accuracy: 0.667\n",
      "Epoch: 450/1000 cost: 0.371460045\n",
      " Training accuracy: 0.700\n",
      " Test accuracy: 0.611\n",
      "Epoch: 460/1000 cost: 0.369880239\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.556\n",
      "Epoch: 470/1000 cost: 0.394161945\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.556\n",
      "Epoch: 480/1000 cost: 0.364306872\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.556\n",
      "Epoch: 490/1000 cost: 0.366211325\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.667\n",
      "Epoch: 500/1000 cost: 0.370329003\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.556\n",
      "Epoch: 510/1000 cost: 0.329603980\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.556\n",
      "Epoch: 520/1000 cost: 0.363272727\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.667\n",
      "Epoch: 530/1000 cost: 0.357857133\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.611\n",
      "Epoch: 540/1000 cost: 0.338225568\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.611\n",
      "Epoch: 550/1000 cost: 0.343919739\n",
      " Training accuracy: 0.800\n",
      " Test accuracy: 0.667\n",
      "Epoch: 560/1000 cost: 0.413940072\n",
      " Training accuracy: 0.800\n",
      " Test accuracy: 0.611\n",
      "Epoch: 570/1000 cost: 0.361196553\n",
      " Training accuracy: 0.800\n",
      " Test accuracy: 0.611\n",
      "Epoch: 580/1000 cost: 0.282896074\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.556\n",
      "Epoch: 590/1000 cost: 0.326758489\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.611\n",
      "Epoch: 600/1000 cost: 0.403622220\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.667\n",
      "Epoch: 610/1000 cost: 0.367018700\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.667\n",
      "Epoch: 620/1000 cost: 0.341980045\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.556\n",
      "Epoch: 630/1000 cost: 0.340703443\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.667\n",
      "Epoch: 640/1000 cost: 0.328072707\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.667\n",
      "Epoch: 650/1000 cost: 0.309617691\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.611\n",
      "Epoch: 660/1000 cost: 0.344842007\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.611\n",
      "Epoch: 670/1000 cost: 0.332698236\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.667\n",
      "Epoch: 680/1000 cost: 0.380503545\n",
      " Training accuracy: 0.800\n",
      " Test accuracy: 0.667\n",
      "Epoch: 690/1000 cost: 0.280142625\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.611\n",
      "Epoch: 700/1000 cost: 0.267455901\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.611\n",
      "Epoch: 710/1000 cost: 0.362288361\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.667\n",
      "Epoch: 720/1000 cost: 0.358635515\n",
      " Training accuracy: 0.800\n",
      " Test accuracy: 0.667\n",
      "Epoch: 730/1000 cost: 0.362987389\n",
      " Training accuracy: 0.700\n",
      " Test accuracy: 0.667\n",
      "Epoch: 740/1000 cost: 0.318262411\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.667\n",
      "Epoch: 750/1000 cost: 0.307763154\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.667\n",
      "Epoch: 760/1000 cost: 0.286715418\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.611\n",
      "Epoch: 770/1000 cost: 0.308130321\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.611\n",
      "Epoch: 780/1000 cost: 0.283916379\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.667\n",
      "Epoch: 790/1000 cost: 0.311522640\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.667\n",
      "Epoch: 800/1000 cost: 0.278072645\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.611\n",
      "Epoch: 810/1000 cost: 0.275094579\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.611\n",
      "Epoch: 820/1000 cost: 0.317040302\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.611\n",
      "Epoch: 830/1000 cost: 0.279060165\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.611\n",
      "Epoch: 840/1000 cost: 0.278167590\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.667\n",
      "Epoch: 850/1000 cost: 0.358885030\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.667\n",
      "Epoch: 860/1000 cost: 0.332068468\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.611\n",
      "Epoch: 870/1000 cost: 0.329314743\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.667\n",
      "Epoch: 880/1000 cost: 0.305011570\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.667\n",
      "Epoch: 890/1000 cost: 0.307910045\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.667\n",
      "Epoch: 900/1000 cost: 0.312113340\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.667\n",
      "Epoch: 910/1000 cost: 0.289214273\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.611\n",
      "Epoch: 920/1000 cost: 0.236031855\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.611\n",
      "Epoch: 930/1000 cost: 0.266551378\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.667\n",
      "Epoch: 940/1000 cost: 0.261112727\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.667\n",
      "Epoch: 950/1000 cost: 0.237386634\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.556\n",
      "Epoch: 960/1000 cost: 0.297203581\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.667\n",
      "Epoch: 970/1000 cost: 0.270228222\n",
      " Training accuracy: 0.700\n",
      " Test accuracy: 0.611\n",
      "Epoch: 980/1000 cost: 0.286420777\n",
      " Training accuracy: 1.000\n",
      " Test accuracy: 0.611\n",
      "Epoch: 990/1000 cost: 0.284095377\n",
      " Training accuracy: 0.900\n",
      " Test accuracy: 0.667\n",
      "Optimization Finished!\n"
     ]
    }
   ],
   "source": [
    "# Launch the graph\n",
    "sess = tf.Session()\n",
    "sess.run(init)\n",
    "\n",
    "# Training cycle\n",
    "for epoch in range(training_epochs):\n",
    "    avg_cost = 0.\n",
    "    num_batch = int(ntrain/batch_size)\n",
    "    # Loop over all batches\n",
    "    for i in range(num_batch): \n",
    "        randidx = np.random.randint(ntrain, size=batch_size)\n",
    "        batch_xs = trainimg[randidx, :]\n",
    "        batch_ys = trainlabel[randidx, :]                \n",
    "        # Fit training using batch data\n",
    "        sess.run(optm, feed_dict={x: batch_xs, y: batch_ys})\n",
    "        # Compute average loss\n",
    "        avg_cost += sess.run(cost, feed_dict={x: batch_xs, y: batch_ys})/num_batch\n",
    "\n",
    "    # Display logs per epoch step\n",
    "    if epoch % display_step == 0:\n",
    "        print (\"Epoch: %03d/%03d cost: %.9f\" % (epoch, training_epochs, avg_cost))\n",
    "        train_acc = sess.run(accr, feed_dict={x: batch_xs, y: batch_ys})\n",
    "        print (\" Training accuracy: %.3f\" % (train_acc))\n",
    "        test_acc = sess.run(accr, feed_dict={x: testimg, y: testlabel})\n",
    "        print (\" Test accuracy: %.3f\" % (test_acc))\n",
    "\n",
    "print (\"Optimization Finished!\")\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.4.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}