10_CNN_fashion_dataset_example.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now that we know the basics of keras lets create an example: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 209,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "<style>\n",
       ".output {\n",
       "    display: flex;\n",
       "    align-items: center;\n",
       "    text-align: center;\n",
       "}\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " TensorFlow Version: 2.2.0\n"
     ]
    }
   ],
   "source": [
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense, Flatten, Conv2D, MaxPooling2D\n",
    "from tensorflow.keras.preprocessing import image\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import tensorflow as tf\n",
    "import random\n",
    "\n",
    "from IPython.display import display, HTML\n",
    "\n",
    "\n",
    "display(HTML(\"\"\"\n",
    "<style>\n",
    ".output {\n",
    "    display: flex;\n",
    "    align-items: center;\n",
    "    text-align: center;\n",
    "}\n",
    "</style>\n",
    "\"\"\"))\n",
    "print(f\" TensorFlow Version: {tf.__version__}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1.- Load the Fashion-MNIST dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x_train shape: (60000, 28, 28) y_train shape: (60000,)\n"
     ]
    }
   ],
   "source": [
    "# Load the fashion-mnist pre-shuffled train data and test data\n",
    "(x_train, y_train), (x_test, y_test) = tf.keras.datasets.fashion_mnist.load_data()\n",
    "\n",
    "print(\"x_train shape:\", x_train.shape, \"y_train shape:\", y_train.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 136,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x_train shape: (55000, 28, 28, 1) y_train shape: (55000, 10)\n",
      "55000 train set\n",
      "10000 test set\n"
     ]
    }
   ],
   "source": [
    "# Print training set shape - note there are 60,000 training data of image size of 28x28, 60,000 train labels)\n",
    "print(\"x_train shape:\", x_train.shape, \"y_train shape:\", y_train.shape)\n",
    "\n",
    "# Print the number of training and test datasets\n",
    "print(x_train.shape[0], 'train set')\n",
    "print(x_test.shape[0], 'test set')\n",
    "\n",
    "# Define the text labels\n",
    "fashion_mnist_labels = [\"T-shirt/top\",  # index 0\n",
    "                        \"Trouser\",      # index 1\n",
    "                        \"Pullover\",     # index 2 \n",
    "                        \"Dress\",        # index 3 \n",
    "                        \"Coat\",         # index 4\n",
    "                        \"Sandal\",       # index 5\n",
    "                        \"Shirt\",        # index 6 \n",
    "                        \"Sneaker\",      # index 7 \n",
    "                        \"Bag\",          # index 8 \n",
    "                        \"Ankle boot\"]   # index 9"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train = x_train.astype('float32') / 255\n",
    "x_test = x_test.astype('float32') / 255"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of train data - 60000\n",
      "Number of test data - 10000\n"
     ]
    }
   ],
   "source": [
    "print(\"Number of train data - \" + str(len(x_train)))\n",
    "print(\"Number of test data - \" + str(len(x_test)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x_train shape: (55000, 28, 28, 1) y_train shape: (55000, 10)\n",
      "55000 train set\n",
      "5000 validation set\n",
      "10000 test set\n"
     ]
    }
   ],
   "source": [
    "# Further break training data into train / validation sets (# put 5000 into validation set and keep remaining 55,000 for train)\n",
    "(x_train, x_valid) = x_train[5000:], x_train[:5000] \n",
    "(y_train, y_valid) = y_train[5000:], y_train[:5000]\n",
    "\n",
    "# Reshape input data from (28, 28) to (28, 28, 1)\n",
    "w, h = 28, 28\n",
    "x_train = x_train.reshape(x_train.shape[0], w, h, 1)\n",
    "x_valid = x_valid.reshape(x_valid.shape[0], w, h, 1)\n",
    "x_test = x_test.reshape(x_test.shape[0], w, h, 1)\n",
    "\n",
    "# One-hot encode the labels\n",
    "y_train = tf.keras.utils.to_categorical(y_train, 10)\n",
    "y_valid = tf.keras.utils.to_categorical(y_valid, 10)\n",
    "y_test = tf.keras.utils.to_categorical(y_test, 10)\n",
    "\n",
    "# Print training set shape\n",
    "print(\"x_train shape:\", x_train.shape, \"y_train shape:\", y_train.shape)\n",
    "\n",
    "# Print the number of training, validation, and test datasets\n",
    "print(x_train.shape[0], 'train set')\n",
    "print(x_valid.shape[0], 'validation set')\n",
    "print(x_test.shape[0], 'test set')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# settings\n",
    "nrows, ncols = 2, 5  # array of sub-plots\n",
    "figsize = [15, 8]     # figure size, inches\n",
    "\n",
    "\n",
    "# create figure (fig), and array of axes (ax)\n",
    "fig, ax = plt.subplots(nrows=nrows, ncols=ncols, figsize=figsize)\n",
    "\n",
    "# plot simple raster image on each sub-plot\n",
    "for i, axi in enumerate(ax.flat):\n",
    "    example = random.randrange(0,x_train.shape[0]) # Select number of example\n",
    "    y_class = np.argmax(y_train[example])          # Get Numerical class out of one-hote encoded vector\n",
    "    label = fashion_mnist_labels[y_class]          # Get Word for label\n",
    "    x_shaped = x_train[example].reshape(x_train.shape[1],x_train.shape[2]) # reshape x\n",
    "    axi.imshow(x_shaped)                           # Plot x\n",
    "    axi.axis('off')                                # Get rid of axis\n",
    "    axi.set_title(f\"{label}\")                      # Set title\n",
    "\n",
    "\n",
    "plt.tight_layout(True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.- Create the Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_8\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_10 (Conv2D)           (None, 28, 28, 64)        320       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_10 (MaxPooling (None, 14, 14, 64)        0         \n",
      "_________________________________________________________________\n",
      "dropout_6 (Dropout)          (None, 14, 14, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_11 (Conv2D)           (None, 14, 14, 32)        8224      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_11 (MaxPooling (None, 7, 7, 32)          0         \n",
      "_________________________________________________________________\n",
      "dropout_7 (Dropout)          (None, 7, 7, 32)          0         \n",
      "_________________________________________________________________\n",
      "flatten_8 (Flatten)          (None, 1568)              0         \n",
      "_________________________________________________________________\n",
      "dense_14 (Dense)             (None, 256)               401664    \n",
      "_________________________________________________________________\n",
      "dropout_8 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_15 (Dense)             (None, 10)                2570      \n",
      "=================================================================\n",
      "Total params: 412,778\n",
      "Trainable params: 412,778\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = tf.keras.Sequential()\n",
    "\n",
    "# Must define the input shape in the first layer of the neural network\n",
    "model.add(tf.keras.layers.Conv2D(filters=64, kernel_size=2, padding='same', activation='relu', input_shape=(28,28,1))) \n",
    "model.add(tf.keras.layers.MaxPooling2D(pool_size=2))\n",
    "model.add(tf.keras.layers.Dropout(0.3))\n",
    "\n",
    "model.add(tf.keras.layers.Conv2D(filters=32, kernel_size=2, padding='same', activation='relu'))\n",
    "model.add(tf.keras.layers.MaxPooling2D(pool_size=2))\n",
    "model.add(tf.keras.layers.Dropout(0.3))\n",
    "\n",
    "model.add(tf.keras.layers.Flatten())\n",
    "model.add(tf.keras.layers.Dense(256, activation='relu'))\n",
    "model.add(tf.keras.layers.Dropout(0.5))\n",
    "model.add(tf.keras.layers.Dense(10, activation='softmax'))\n",
    "\n",
    "# Take a look at the model summary\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.- Compile Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss='categorical_crossentropy',\n",
    "             optimizer='adam',\n",
    "             metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5.- Fit Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/4\n",
      "860/860 [==============================] - 92s 107ms/step - loss: 0.2224 - accuracy: 0.9184 - val_loss: 0.2032 - val_accuracy: 0.9228\n",
      "Epoch 2/4\n",
      "860/860 [==============================] - 90s 104ms/step - loss: 0.2186 - accuracy: 0.9189 - val_loss: 0.2045 - val_accuracy: 0.9198\n"
     ]
    }
   ],
   "source": [
    "training = model.fit(x_train,\n",
    "         y_train,\n",
    "         batch_size=64,\n",
    "         epochs=4,\n",
    "         validation_data=(x_valid, y_valid),\n",
    "         verbose = 1,\n",
    "         callbacks=[tf.keras.callbacks.EarlyStopping()]\n",
    "         )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Notice that since we saved the Trainment of the model in a variable we can acces it's history"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 193,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'loss': [0.2224215418100357, 0.2185741811990738],\n",
       " 'accuracy': [0.9184363484382629, 0.9188727140426636],\n",
       " 'val_loss': [0.20323315262794495, 0.20447351038455963],\n",
       " 'val_accuracy': [0.9228000044822693, 0.9197999835014343]}"
      ]
     },
     "execution_count": 193,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "training.history"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can use this information to create plots"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 206,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x432 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(15,6))\n",
    "### Plot accuracy ###\n",
    "ax[0].set_title(\"Accuracy\")\n",
    "ax[0].plot(training.history[\"val_accuracy\"], label = \"Validation Accuracy\")\n",
    "ax[0].plot(training.history[\"accuracy\"],     label =\"Training Accuracy\")\n",
    "ax[0].legend()\n",
    "\n",
    "### Plot loss ###\n",
    "ax[1].set_title(\"Loss\")\n",
    "ax[1].plot(training.history[\"val_loss\"], label =\"Validation Loss\")\n",
    "ax[1].plot(training.history[\"loss\"],     label =\"Training Loss\")\n",
    "ax[1].legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can use the method evaluate() to check the test set. evaluate() will return the loss plus the metrics we defined"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 [==============================] - 5s 15ms/step - loss: 0.2252 - accuracy: 0.9158\n"
     ]
    }
   ],
   "source": [
    "loss_test, accuracy_test = model.evaluate(x_test ,y_test)\n",
    "print(\"loss in test set: {loss_test}, Accuracy in test set: {acuraccy_test}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_hat = model.predict(x_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(10000, 28, 28, 1)\n",
      "(28, 28, 1)\n"
     ]
    }
   ],
   "source": [
    "# Notice that: tensorflow expect a tensro of shape n,28,28,1\n",
    "# when we slice one example we loose  n so we have to add an axis to use \n",
    "# mode.predic() for one example\n",
    "print(x_test.shape)\n",
    "print(x_test[1].shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_predict_1 = model.predict(x_test[1][np.newaxis,...])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "def plot_prediction(x,y):\n",
    "    example = random.randrange(0,x.shape[0])                 # Select number of example\n",
    "\n",
    "    y_predict_1 = model.predict(x[example][np.newaxis,...])  # Make perdiction \n",
    "    y_predict_1  = np.argmax(y_predict_1)                    #  Get Numerical class out of one-hote encoded prediction\n",
    "    y_class_true = np.argmax(y[example])                     # Get Numerical class out of one-hote encoded real label\n",
    "    label_predicted = fashion_mnist_labels[y_predict_1]      # Get Word for prediction label\n",
    "    label_true = fashion_mnist_labels[y_class_true]          # Get Word for real label\n",
    "    x_shaped = x[example].reshape(x.shape[1],x.shape[2])     # Reshape x into an (pixle,pixel) Matrix\n",
    "    plt.imshow(x_shaped)                                     # Plot Matrix\n",
    "    plt.axis('off')\n",
    "    if label_true == label_predicted:                        # Check if labels are correct\n",
    "        status =\"prediction correct\"                         # If labels are equal, labels are correct\n",
    "    else:\n",
    "        status =\"prediction incorrect\"                       # Else they are incorrect\n",
    "\n",
    "    # Set title \n",
    "    plt.title(f\"Real label:{label_true} | Predicton {label_predicted} | {status}\")\n",
    "    plt.show()\n",
    "\n",
    "plot_prediction(x_test,y_test)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "machine_learning_env",
   "language": "python",
   "name": "machine_learning_env"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}