DigitRecognizer.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## In the Name of God\n",
    "<h2>Digit recogniser using Hoda Dataset\n",
    "<br>Teacher: Dr. Mohsen Hooshmand\n",
    "<br>Advanced Algorithm Course Project\n",
    "<br>Reza Khaleghi - 984118</h2>\n",
    "<br><br>Persian digits dataset: <a href=\"https://github.com/amir-saniyan/HodaDatasetReader\" target=\"_blank\">Hoda Dataset Reader</a>\n",
    "<br><b>Note: This dataset is free of charge for research purposes and non commercial uses only."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Importing required packages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from matplotlib import pyplot as plt\n",
    "from HodaDatasetReader import read_hoda_cdb, read_hoda_dataset\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense, Conv2D, Dropout, Flatten, MaxPooling2D\n",
    "from sklearn.metrics import classification_report"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Reading from Hoda dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "################################################################################\n",
      "\n",
      "Reading Train 60000.cdb ...\n",
      "Reading Test 20000.cdb ...\n",
      "type(train_images):  <class 'list'>\n",
      "len(train_images):  60000\n",
      "\n",
      "type(train_labels):  <class 'list'>\n",
      "len(train_labels):  60000\n",
      "\n",
      "----------------------------------------\n",
      "\n",
      "type(train_images[ 0 ]): <class 'numpy.ndarray'>\n",
      "train_images[ 0 ].dtype: uint8\n",
      "train_images[ 0 ].min(): 0\n",
      "train_images[ 0 ].max(): 255\n",
      "train_images[ 0 ].shape = (HEIGHT, WIDTH): (27, 20)\n",
      "\n",
      "type(train_labels[ 0 ]): <class 'int'>\n",
      "train_labels[ 0 ]: 6\n",
      "\n",
      "----------------------------------------\n",
      "\n",
      "type(train_images[ 1 ]): <class 'numpy.ndarray'>\n",
      "train_images[ 1 ].dtype: uint8\n",
      "train_images[ 1 ].min(): 0\n",
      "train_images[ 1 ].max(): 255\n",
      "train_images[ 1 ].shape = (HEIGHT, WIDTH): (20, 21)\n",
      "\n",
      "type(train_labels[ 1 ]): <class 'int'>\n",
      "train_labels[ 1 ]: 5\n",
      "\n",
      "----------------------------------------\n",
      "\n",
      "type(train_images[ 2 ]): <class 'numpy.ndarray'>\n",
      "train_images[ 2 ].dtype: uint8\n",
      "train_images[ 2 ].min(): 0\n",
      "train_images[ 2 ].max(): 255\n",
      "train_images[ 2 ].shape = (HEIGHT, WIDTH): (10, 15)\n",
      "\n",
      "type(train_labels[ 2 ]): <class 'int'>\n",
      "train_labels[ 2 ]: 0\n",
      "\n",
      "----------------------------------------\n",
      "\n",
      "type(train_images[ 3 ]): <class 'numpy.ndarray'>\n",
      "train_images[ 3 ].dtype: uint8\n",
      "train_images[ 3 ].min(): 0\n",
      "train_images[ 3 ].max(): 255\n",
      "train_images[ 3 ].shape = (HEIGHT, WIDTH): (36, 17)\n",
      "\n",
      "type(train_labels[ 3 ]): <class 'int'>\n",
      "train_labels[ 3 ]: 2\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x288 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "################################################################################\n",
      "\n",
      "Reading train dataset (Train 60000.cdb)...\n",
      "Reading test dataset (Test 20000.cdb)...\n",
      "Done\n"
     ]
    }
   ],
   "source": [
    "print('################################################################################')\n",
    "print()\n",
    "\n",
    "print('Reading Train 60000.cdb ...')\n",
    "train_images, train_labels = read_hoda_cdb('./DigitDB/Train 60000.cdb')\n",
    "\n",
    "print('Reading Test 20000.cdb ...')\n",
    "test_images, test_labels = read_hoda_cdb('./DigitDB/Test 20000.cdb')\n",
    "\n",
    "# ******************************************************************************\n",
    "\n",
    "print('type(train_images): ', type(train_images))\n",
    "print('len(train_images): ', len(train_images))\n",
    "print()\n",
    "\n",
    "print('type(train_labels): ', type(train_labels))\n",
    "print('len(train_labels): ', len(train_labels))\n",
    "print()\n",
    "\n",
    "fig = plt.figure(figsize=(15, 4))\n",
    "for i in range(4):\n",
    "\n",
    "    print('----------------------------------------')\n",
    "    print()\n",
    "\n",
    "    print('type(train_images[', i, ']):', type(train_images[i]))\n",
    "    print('train_images[', i, '].dtype:', train_images[i].dtype)\n",
    "    print('train_images[', i, '].min():', train_images[i].min())\n",
    "    print('train_images[', i, '].max():', train_images[i].max())\n",
    "    print('train_images[', i, '].shape = (HEIGHT, WIDTH):', train_images[i].shape)\n",
    "    print()\n",
    "\n",
    "    print('type(train_labels[', i, ']):', type(train_labels[i]))\n",
    "    print('train_labels[', i, ']:', train_labels[i])\n",
    "    print()\n",
    "\n",
    "    fig.add_subplot(1, 4, i + 1)\n",
    "    plt.title('train_labels[' + str(i) + '] = ' + str(train_labels[i]))\n",
    "    plt.imshow(train_images[i], cmap='gray')\n",
    "\n",
    "plt.show()\n",
    "\n",
    "print('################################################################################')\n",
    "print()\n",
    "\n",
    "print('Reading train dataset (Train 60000.cdb)...')\n",
    "x_train, y_train = read_hoda_dataset(dataset_path='./DigitDB/Train 60000.cdb',\n",
    "                                images_height=32,\n",
    "                                images_width=32,\n",
    "                                one_hot=False,\n",
    "                                reshape=True)\n",
    "\n",
    "print('Reading test dataset (Test 20000.cdb)...')\n",
    "x_test, y_test = read_hoda_dataset(dataset_path='./DigitDB/Test 20000.cdb',\n",
    "                              images_height=32,\n",
    "                              images_width=32,\n",
    "                              one_hot=True,\n",
    "                              reshape=False)\n",
    "print(\"Done\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<b>Checking if the dataset imported correctly"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "7.0\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x246fdad8a08>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "image_index = 7777 # You may select anything up to 60,000\n",
    "print(y_train[image_index]) # The label is 7\n",
    "plt.imshow(train_images[image_index], cmap='gray')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x_train shape: (60000, 32, 32, 1)\n",
      "Number of images in x_train 60000\n",
      "Number of images in x_test 20000\n"
     ]
    }
   ],
   "source": [
    "# Reshaping the array to 4-dims so that it can work with the Keras API\n",
    "x_train = x_train.reshape(x_train.shape[0], 32, 32, 1)\n",
    "x_test = x_test.reshape(x_test.shape[0], 32, 32, 1)\n",
    "input_shape = (32, 32, 1)\n",
    "# Making sure that the values are float so that we can get decimal points after division\n",
    "x_train = x_train.astype('float32')\n",
    "x_test = x_test.astype('float32')\n",
    "# Normalizing the RGB codes by dividing it to the max RGB value. changing the range from 0-255 to 0-1\n",
    "x_train /= 255\n",
    "x_test /= 255\n",
    "\n",
    "print('x_train shape:', x_train.shape)\n",
    "print('Number of images in x_train', x_train.shape[0])\n",
    "print('Number of images in x_test', x_test.shape[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Creating a Sequential Model and adding the layers\n",
    "model = []\n",
    "model.append(Sequential())\n",
    "model.append(Sequential())\n",
    "model.append(Sequential())\n",
    "model.append(Sequential())\n",
    "\n",
    "dropout = [0.1,0.2,0.5,0.8]\n",
    "for i in range(4):\n",
    "    model[i].add(Conv2D(28, kernel_size=(3,3), input_shape=input_shape))\n",
    "    model[i].add(MaxPooling2D(pool_size=(2, 2)))\n",
    "    model[i].add(Flatten()) # Flattening the 2D arrays for fully connected layers\n",
    "    model[i].add(Dense(128, activation=tf.nn.relu))\n",
    "    model[i].add(Dropout(dropout[i]))\n",
    "    model[i].add(Dense(10,activation=tf.nn.softmax))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<b>Fitting the model with 10 epochs.\n",
    "<br>Note: Increasing epochs will increase accuracy but also increases processing time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "1875/1875 [==============================] - 14s 8ms/step - loss: 0.3613 - accuracy: 0.8970\n",
      "Epoch 2/10\n",
      "1875/1875 [==============================] - 14s 7ms/step - loss: 0.0959 - accuracy: 0.9720\n",
      "Epoch 3/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0601 - accuracy: 0.9829\n",
      "Epoch 4/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0452 - accuracy: 0.9863\n",
      "Epoch 5/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0374 - accuracy: 0.9882\n",
      "Epoch 6/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0306 - accuracy: 0.9902\n",
      "Epoch 7/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0283 - accuracy: 0.9909\n",
      "Epoch 8/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0232 - accuracy: 0.9924\n",
      "Epoch 9/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0216 - accuracy: 0.9930\n",
      "Epoch 10/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0194 - accuracy: 0.9936\n",
      "Epoch 1/10\n",
      "1875/1875 [==============================] - 14s 7ms/step - loss: 0.3620 - accuracy: 0.9012\n",
      "Epoch 2/10\n",
      "1875/1875 [==============================] - 14s 7ms/step - loss: 0.0990 - accuracy: 0.9712\n",
      "Epoch 3/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0681 - accuracy: 0.9808\n",
      "Epoch 4/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0535 - accuracy: 0.9844\n",
      "Epoch 5/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0454 - accuracy: 0.9870\n",
      "Epoch 6/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0373 - accuracy: 0.9885\n",
      "Epoch 7/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0353 - accuracy: 0.9889\n",
      "Epoch 8/10\n",
      "1875/1875 [==============================] - 12s 7ms/step - loss: 0.0299 - accuracy: 0.9906\n",
      "Epoch 9/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0284 - accuracy: 0.9905\n",
      "Epoch 10/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0238 - accuracy: 0.9922\n",
      "Epoch 1/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.3851 - accuracy: 0.8893\n",
      "Epoch 2/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.1267 - accuracy: 0.9637\n",
      "Epoch 3/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.1000 - accuracy: 0.9722\n",
      "Epoch 4/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0855 - accuracy: 0.9757\n",
      "Epoch 5/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0773 - accuracy: 0.9784\n",
      "Epoch 6/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0735 - accuracy: 0.9788\n",
      "Epoch 7/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0676 - accuracy: 0.9801\n",
      "Epoch 8/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0627 - accuracy: 0.9817\n",
      "Epoch 9/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0612 - accuracy: 0.9820\n",
      "Epoch 10/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.0574 - accuracy: 0.9827\n",
      "Epoch 1/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.5806 - accuracy: 0.8242\n",
      "Epoch 2/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.2667 - accuracy: 0.9230\n",
      "Epoch 3/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.2393 - accuracy: 0.9290\n",
      "Epoch 4/10\n",
      "1875/1875 [==============================] - 14s 7ms/step - loss: 0.2210 - accuracy: 0.9350\n",
      "Epoch 5/10\n",
      "1875/1875 [==============================] - 14s 7ms/step - loss: 0.2141 - accuracy: 0.9360\n",
      "Epoch 6/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.2072 - accuracy: 0.9378\n",
      "Epoch 7/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.2054 - accuracy: 0.9375\n",
      "Epoch 8/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.1986 - accuracy: 0.9415\n",
      "Epoch 9/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.1940 - accuracy: 0.9425\n",
      "Epoch 10/10\n",
      "1875/1875 [==============================] - 13s 7ms/step - loss: 0.1882 - accuracy: 0.9435\n"
     ]
    }
   ],
   "source": [
    "for i in range(4):\n",
    "    model[i].compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
    "    model[i].fit(x=x_train,y=y_train, epochs=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "lst = []\n",
    "for i in y_test:\n",
    "    for j in range(10):\n",
    "        if i[j] == 1:\n",
    "            lst.append(j)\n",
    "final_y_test = np.array(lst)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<b>Change the \"image_index\" variable amount in the range of test to show the predicted (here the test set range is between 0-19999)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicted label: 6 \n",
      "With the probability of 0.9999989\n",
      "Predicted label: 6 \n",
      "With the probability of 0.9999999\n",
      "Predicted label: 6 \n",
      "With the probability of 0.9999665\n",
      "Predicted label: 6 \n",
      "With the probability of 0.99997735\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "image_index = 12111\n",
    "plt.imshow(x_test[image_index].reshape(32, 32),cmap='Greys')\n",
    "for i in range(4):\n",
    "    pred = model[i].predict(x_test[image_index].reshape(1, 32, 32, 1))\n",
    "    # Maximum probability\n",
    "    print(\"Predicted label:\",pred.argmax(),\"\\nWith the probability of\",pred[0][pred.argmax()])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model 0\n",
      "157/157 [==============================] - 1s 7ms/step - loss: 0.0791 - accuracy: 0.9809\n",
      "WARNING:tensorflow:5 out of the last 5 calls to <function Model.make_predict_function.<locals>.predict_function at 0x000002468C57ED38> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/tutorials/customization/performance#python_or_tensor_args and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
      "+++++++++++++++++++++++++++++++++++++++++++++++++++++++\n",
      "Model 1\n",
      "157/157 [==============================] - 1s 7ms/step - loss: 0.0805 - accuracy: 0.9791\n",
      "+++++++++++++++++++++++++++++++++++++++++++++++++++++++\n",
      "Model 2\n",
      "157/157 [==============================] - 1s 8ms/step - loss: 0.0864 - accuracy: 0.9758\n",
      "+++++++++++++++++++++++++++++++++++++++++++++++++++++++\n",
      "Model 3\n",
      "157/157 [==============================] - 1s 8ms/step - loss: 0.1607 - accuracy: 0.9557\n",
      "+++++++++++++++++++++++++++++++++++++++++++++++++++++++\n"
     ]
    }
   ],
   "source": [
    "y_pred = []\n",
    "for i in range(4):\n",
    "    print(\"Model\",i)\n",
    "    results = model[i].evaluate(x_test,final_y_test,batch_size=128)\n",
    "    y_pred.append(model[i].predict(x_test))\n",
    "    print(\"+++++++++++++++++++++++++++++++++++++++++++++++++++++++\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       1.00      0.99      0.99      2000\n",
      "           1       0.97      1.00      0.98      2000\n",
      "           2       0.94      0.99      0.96      2000\n",
      "           3       0.98      0.94      0.96      2000\n",
      "           4       0.97      0.98      0.97      2000\n",
      "           5       0.98      0.99      0.99      2000\n",
      "           6       0.99      0.98      0.98      2000\n",
      "           7       0.99      0.99      0.99      2000\n",
      "           8       0.99      0.99      0.99      2000\n",
      "           9       0.99      0.97      0.98      2000\n",
      "\n",
      "    accuracy                           0.98     20000\n",
      "   macro avg       0.98      0.98      0.98     20000\n",
      "weighted avg       0.98      0.98      0.98     20000\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.99      0.99      0.99      2000\n",
      "           1       0.97      1.00      0.98      2000\n",
      "           2       0.94      0.98      0.96      2000\n",
      "           3       0.97      0.94      0.95      2000\n",
      "           4       0.98      0.96      0.97      2000\n",
      "           5       0.99      0.98      0.99      2000\n",
      "           6       0.98      0.98      0.98      2000\n",
      "           7       0.99      0.99      0.99      2000\n",
      "           8       0.99      0.99      0.99      2000\n",
      "           9       0.99      0.97      0.98      2000\n",
      "\n",
      "    accuracy                           0.98     20000\n",
      "   macro avg       0.98      0.98      0.98     20000\n",
      "weighted avg       0.98      0.98      0.98     20000\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.99      0.99      0.99      2000\n",
      "           1       0.97      1.00      0.98      2000\n",
      "           2       0.95      0.96      0.96      2000\n",
      "           3       0.96      0.94      0.95      2000\n",
      "           4       0.96      0.96      0.96      2000\n",
      "           5       0.98      0.99      0.98      2000\n",
      "           6       0.97      0.98      0.98      2000\n",
      "           7       0.99      0.98      0.99      2000\n",
      "           8       0.99      0.99      0.99      2000\n",
      "           9       0.98      0.96      0.97      2000\n",
      "\n",
      "    accuracy                           0.98     20000\n",
      "   macro avg       0.98      0.98      0.98     20000\n",
      "weighted avg       0.98      0.98      0.98     20000\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.99      0.98      0.98      2000\n",
      "           1       0.96      0.99      0.97      2000\n",
      "           2       0.91      0.93      0.92      2000\n",
      "           3       0.91      0.90      0.91      2000\n",
      "           4       0.93      0.93      0.93      2000\n",
      "           5       0.97      0.98      0.98      2000\n",
      "           6       0.96      0.95      0.96      2000\n",
      "           7       0.98      0.97      0.98      2000\n",
      "           8       0.99      0.97      0.98      2000\n",
      "           9       0.96      0.96      0.96      2000\n",
      "\n",
      "    accuracy                           0.96     20000\n",
      "   macro avg       0.96      0.96      0.96     20000\n",
      "weighted avg       0.96      0.96      0.96     20000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "for j in range(4):\n",
    "    converted_y_pred = []\n",
    "    for i in y_pred[j]:\n",
    "        converted_y_pred.append(np.argmax(i))\n",
    "    final_y_pred = np.array(converted_y_pred)\n",
    "    print(classification_report(final_y_test,final_y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}