2_classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Classification MNIST\n",
    "in MNIST dataset, set of 70,000 small image of digits handwritten."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.datasets import fetch_openml\n",
    "\n",
    "mnist = fetch_openml('mnist_784', version=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['data', 'target', 'frame', 'categories', 'feature_names', 'target_names', 'DESCR', 'details', 'url'])"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mnist.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((70000, 784), (70000,))"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X, y = mnist['data'], mnist['target']\n",
    "X.shape, y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pylab as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'5'"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = np.array(X)\n",
    "y = np.array(y, dtype=np.uint8)  #dtype=np.uint8 change to integer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(784,)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X[0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(X[44].reshape(28,28), cmap='binary')\n",
    "plt.title(y[44])\n",
    "plt.axis('off')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Normalized data (between 0 and 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "255.0\n",
      "1.0\n"
     ]
    }
   ],
   "source": [
    "print(X[0].max())\n",
    "X = X/255\n",
    "print(X[0].max())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Spilt datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((60000, 784), (10000, 784))"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train, x_test, y_train, y_test = X[:60000], X[60000:], y[:60000], y[60000:]\n",
    "x_train.shape, x_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### JUST Training a Binary Classifier\n",
    "* number 5 and not-5\n",
    "* set labels betwwen True and False then *1 for 1 and 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1, 0, 0, ..., 1, 0, 0])"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(y_train == 5)*1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0])"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train_5 = (y_train == 5)*1\n",
    "y_test_5 = (y_test == 5)*1\n",
    "y_train_5[:20]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "pick a classifier and train it. A good place to start is with a **Stochastic Gradient Descent (SGD)**, this classifier has advantage of being capable of handling very large datasets efficienly."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SGDClassifier(random_state=42)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.linear_model import SGDClassifier\n",
    "\n",
    "sgd_clf = SGDClassifier(random_state=42)\n",
    "sgd_clf.fit(x_train, y_train_5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### expand_dims()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "if we predict a single value the model what 2D array , we must change x_test[index] to 2D array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_test[0].ndim"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.expand_dims(x_test[0], axis=0).ndim"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[1 4 3]\n",
      " [2 3 4]] \n",
      "\n",
      "[[[1 4 3]\n",
      "  [2 3 4]]] \n",
      "\n",
      "[[[1 4 3]]\n",
      "\n",
      " [[2 3 4]]]\n"
     ]
    }
   ],
   "source": [
    "z = np.array([[1,4,3],[2,3,4]])\n",
    "print(z,'\\n')\n",
    "print(np.expand_dims(z, axis=(0)),'\\n')\n",
    "print(np.expand_dims(z, axis=1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0])"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sgd_clf.predict(np.expand_dims(x_test[0], axis=0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ True,  True,  True,  True,  True,  True,  True,  True,  True,\n",
       "        True,  True,  True,  True,  True,  True,  True,  True, False,\n",
       "        True,  True,  True,  True,  True, False,  True,  True,  True,\n",
       "        True,  True,  True,  True,  True,  True,  True,  True,  True,\n",
       "       False,  True,  True,  True,  True,  True,  True,  True,  True,\n",
       "        True,  True,  True,  True,  True])"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred_5 = sgd_clf.predict(x_test)\n",
    "pred_5[:1000:20] == y_test_5[:1000:20]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Performance Measure\n",
    "* using Cross-Validation\n",
    "* Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.95110978, 0.9639964 , 0.97329733])"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.model_selection import cross_val_score\n",
    "\n",
    "cross_val_score(sgd_clf, x_test, y_test_5, cv=3, scoring='accuracy')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Confusion Matrix from train set by using *Cross val predict* because it's return prediction classes and we don't want to touch test set *remember test set must be on the last moodel*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import cross_val_predict\n",
    "\n",
    "y_train_pred = cross_val_predict(sgd_clf, x_train, y_train_5, cv=3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[53613,   966],\n",
       "       [  840,  4581]], dtype=int64)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "cm = confusion_matrix(y_train_5, y_train_pred)\n",
    "cm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sn\n",
    "import pandas as pd\n",
    "\n",
    "cm_df = pd.DataFrame(cm,columns=['False','True (5)'],index=['False','True (5)'])\n",
    "sn.heatmap(cm_df,annot=True,fmt='d')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> each row represents a n acctual class, while each column represents a predict class\n",
    "\n",
    "> 1st row considers non-5 (the negative class):53613 of them correctly called **True Negative** , while the remaining 966 were wrong class as 5 **False Positive**\n",
    "\n",
    "> 2nd row considers 5-class (the positive class):840 were wrong class called **False Negative** , and 4581 remaining were correctly as 5s **True Positive** "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Precision/Recall"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The confusion matrix give a lot of information, but sometimes you may prefer a more concise metrix, called **Precision of the classifier**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**precision** = TP / (TP + FP)\n",
    "\n",
    "precision = 100% if don't have False Positive\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**recall** = TP / (TP + FN)\n",
    "\n",
    "recall or sensitivity or true positive rate (TPR) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "precision : 0.8258518117901569\n",
      "recall : 0.8450470392916436\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import precision_score, recall_score\n",
    "\n",
    "print('precision :',precision_score(y_train_5, y_train_pred))\n",
    "print('recall :',recall_score(y_train_5, y_train_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">Precision: how many of the returned hits were true positive i.e. how many of the found were correct hits. >> 82%\n",
    "\n",
    "> Recall literally is how many of the true positives were recalled (found), i.e. how many of the correct hits were also found. >> 84%\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "it's often convenient to combine precision and recall into single metric called **F1 Score** , the F1 score high if precision and recall are high.\n",
    "\n",
    "F1 = 2(pr*re)/(pr+re)\n",
    "\n",
    "You can have a precision of 1 (so when you say it's positive, it's actutally positive) but still have a very low recall (you predicted 3 good positives but forgot 15 others). Or you can have a good recall and a bad precision.\n",
    "then using F1 score to check it"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "F1 score : 0.8353391684901532\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import f1_score\n",
    "\n",
    "print('F1 score :',f1_score(y_train_5, y_train_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "but F1 score does not everything, some contexts you mostly care about precision, and other contexts you really care about recall.\n",
    "\n",
    "for examples: \n",
    "* if you detect video are safe for kids, you must keep only safe one i.e. high precision. \n",
    "* on the other hand suppose to detect shoplifters in surveillance images, it's must high recall (the security will get few false alerts but almost all shoplifters will get caught) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Precision/Recall Trad-off** : If you increase precision, it will reduce recall and vice versa."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "y_score: [1.36162725] predict: [1]\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "index = 35\n",
    "y_scores = sgd_clf.decision_function([x_train[index]])\n",
    "plt.imshow(x_train[index].reshape(28,28))\n",
    "print('y_score:',y_scores,'predict:', sgd_clf.predict([x_train[index]]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> *Decision function* returns array of shape (n_samples, n_classes) with filled confidence score, which is basically a signed distance of sample from hyperplane of your model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ True])"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "threshold = 1\n",
    "y_some_digit_pred = (y_scores > threshold)\n",
    "y_some_digit_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 1.65073246, -5.94781643, -8.22853161, ...,  2.62249961,\n",
       "       -2.09502974, -2.54393863])"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_scores = cross_val_predict(sgd_clf, x_train, y_train_5, cv=3, method='decision_function')\n",
    "y_scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "precision: [0.09080706 0.09079183 0.09079335] \n",
      "recalls: [1.         0.99981553 0.99981553] \n",
      "thresholds: [-17.5511157  -17.54794245 -17.54281162]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import precision_recall_curve\n",
    "\n",
    "precisions, recalls, thresholds = precision_recall_curve(y_train_5, y_scores)\n",
    "print('precision:',precisions[:3],'\\nrecalls:',recalls[:3],'\\nthresholds:',thresholds[:3])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((59698,), (59698,), (59698,))"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "precisions[:-1].shape, recalls[:-1].shape, thresholds.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def plot_precision_recall_vs_threshold(precisions, recalls, thresholds):\n",
    "    plt.plot(thresholds, precisions[:-1], 'b--', label='Precision')\n",
    "    plt.plot(thresholds, recalls[:-1], 'g-', label='Recall')\n",
    "    plt.xlabel('Threshold')\n",
    "    plt.grid()\n",
    "    plt.legend()\n",
    "\n",
    "plot_precision_recall_vs_threshold(precisions,recalls,thresholds)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> when we set thresholds high precision will increase and recalls going down"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([False, False, False, ...,  True,  True,  True])"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "precisions >= .9"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "55102"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.argmax(precisions >= .9)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8999347400478573"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "precisions[55101]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.6060137354660038"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "threshold_90_precision = thresholds[np.argmax(precisions >= .9)]\n",
    "threshold_90_precision # thresholds that get 90 precision"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1, 0, 0, ..., 1, 0, 0])"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train_pred_90 = (y_scores >= threshold_90_precision)*1\n",
    "y_train_pred_90"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9001305483028721\n",
      "0.7631433314886552\n"
     ]
    }
   ],
   "source": [
    "print(precision_score(y_train_5, y_train_pred_90))\n",
    "print(recall_score(y_train_5, y_train_pred_90))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> you have a 90% precision classifier!, but wait a high-precision not useful if its recall is too low!\n",
    "\n",
    "if someone says, \"Let's reach 99% precision,\"you shold ask,\"At what recall?\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### The ROC curve\n",
    "\n",
    "The receiver operating characteristic : another common tool used with binary classifiers. similar to precision/recall but *plot TPR(anoter name for recall) vs false positive rate (FPR)*\n",
    "\n",
    "ROC Curves are used to see how well your classifier can separate positive and negative examples and to identify the best threshold for separating them"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_curve\n",
    "\n",
    "FPR, TPR , thresholds = roc_curve(y_train_5, y_scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def plot_roc_curve(FPR,TPR,label=None) :\n",
    "    plt.plot(FPR, TPR, linewidth=2, label=label)\n",
    "    plt.plot([0,1],[0,1], 'k--') # Dashed diagonal\n",
    "    plt.xlabel('False Positive Rate')\n",
    "    plt.ylabel('True Positive Rate (recalls)')\n",
    "    plt.grid()\n",
    "    plt.legend()\n",
    "\n",
    "plot_roc_curve(FPR,TPR,'FPR x TPR')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> *area under the curve (AUC)* .a perfect classifier will have a ROC AUC equal to 1 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9704522882419195"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import roc_auc_score\n",
    "\n",
    "roc_auc_score(y_train_5, y_scores)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> how to decied which one to use. you should prefer the PR curve whenever when you care about *False Positive* . other wise use ROC curve"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Use RandomForestClassifier to compare with SGDClassifier, but it does not have a decision_function(). instead it has a *predict_proba()* \n",
    "\n",
    "**Predict Proba** : retuen probability that the given instance belongs to the given class (e.g., 80% chance that represent a 5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.11, 0.89],\n",
       "       [0.99, 0.01],\n",
       "       [0.96, 0.04],\n",
       "       ...,\n",
       "       [0.02, 0.98],\n",
       "       [0.92, 0.08],\n",
       "       [0.94, 0.06]])"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "forest_clf = RandomForestClassifier(random_state=42)\n",
    "y_probas_forest = cross_val_predict(forest_clf, x_train, y_train_5, cv=3,method='predict_proba')\n",
    "y_probas_forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.89, 0.01, 0.04, ..., 0.98, 0.08, 0.06])"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_probas_forest[:,1] # sure it 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_scores_forest = y_probas_forest[:,1]\n",
    "FPR_forest, TPR_forest, threshold_forest = roc_curve(y_train_5, y_scores_forest)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEGCAYAAABo25JHAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAAsTAAALEwEAmpwYAABDzklEQVR4nO3de5yM5fvA8c9lF7tYK8eUYyVZhJwSSVFS3846+kpFUvhV6osOxJdOUs6KxKbDVykVkkRWinJqiZWSkCLWcZdde7p+fzwzu0trd5Z9dnZ2rvfrNa+Ze57Tde8w1zz3/Tz3LaqKMcaY4FXC3wEYY4zxL0sExhgT5CwRGGNMkLNEYIwxQc4SgTHGBLlQfweQX5UrV9Y6deqc1rZHjx6lbNmyBRtQEWd1Dg5W5+BwJnVeu3ZtvKpWyWlZwCWCOnXqsGbNmtPaNiYmhg4dOhRsQEWc1Tk4WJ2Dw5nUWUR2nGqZNQ0ZY0yQs0RgjDFBzhKBMcYEOUsExhgT5CwRGGNMkHMtEYjIdBHZKyIbT7FcRGS8iGwVkQ0icolbsRhjjDk1N88IooFrc1neBajnefQGXncxFmOMMafg2n0EqvqNiNTJZZWbgJnqjIP9vYhUEJHqqrrbrZgKm6pyPC2D5NR0klMzOJqSxrHj6RxNSePo8TSOpqRz7Hgax1LSScvIIDVdSU3PIC1dSc1wntPSM0hJV9IzMlAFzdw3KM4bmu146lkGzvt//53M7L/WedZTZ7vsrzPXz17O2o+3nHP9TiqjOS474fWp1jlxR2fk0OEkJv+88sx2EmCszoFJFY4ehdJhUDIUUtMgPh7OOgvCSkNKCuz8A6qfnUZG4l9Uq1wBN26d8OcNZecCf2Qr7/K8949EICK9cc4aqFatGjExMad1wMTExHxtq6ocTYXDKcrh40piqpKUqhxLg6Opmvk4lgpJaUpSupKcBslpSkoGpGWcVpgFb0+xya2+O3jA3xEUPqtzsfFHtv+yKfG/sXHmODKOHaLNgCmn/f2XG38mAsnhvRx/C6rqVGAqQIsWLfR076zL7a689Axl695Eftx5kB9+P8D6Pw6x80ASaRln9vO0VGgJwkJLEFYyhLKlQylbOoQypUIpW8pTLhVKeKkQSoYIoSElKFnCeQ4NEUqFlCDUWy4hiIAgmX85AUQk8w8pQuY64nlzc9xmohpGedbNWuYtc0JZstbLdizvMq/sH5yc9ClKtqXZl8mJK+Vv/XyKjY2ladOmZ7CHwFPQdU5Ldz6DkBCnHL/f+YVarpxT3rkTIsrDWRUgLQ2++goaN4YaNeDwYZj9EVzZAc4/39n2/ffh+uuccux6+OQTuK8H1K0LW7fCtGnQpw/UqeMsHzECxo+Dc8+F776D8RNg7BioXh0+/RTGjIHnn99Iu3aNePddePNNWLIEQkPhpZfgiy9g2TIn1tGvOvv4ZI5TfnYIrPoBFi1yys88A7t3w/TpTnngIKcOU95wyo8/DqmpMHGiU+7X3znO2DFOuU8fOOdcGDrEKT/6GFSoAMOHOeWnnoImTeGuO53ySy9D61Zw5ZXO327ZN1C7FlxwgVM+eBDKRwKazLRxr/DeOxOpcFYl/jPqNaqfe5Yrd1P7MxHsAmpmK9cA/irsINbuOMC89btZuHEPe44k/2N5ROlQqkSUpnJEaSqWKUVEWCgRYSWpUMZ5RIaXpHx4ScqHhVKudEnKhYVSpmQI4aVCKBVSghIlzuQr7cxFHvqVDk3O8WsMhS1pZwitz6vk7zBOkJoKx45BRASUKAF798Kff0LTpk4CXL8etmyBO+5w1v/yS+cLsm9fp/zZZ7B9Ozz6qFP+4ANn+wEDnPInU88hYVslnnzSKT/1lHO8ceOc8h13OF9OU6c65RYtnGNPm+aU69WDyy/P+jI891x47DHo0QOqVoXwhtC/P4wa5Sy/rB48/bTzhZ2QAO36weDBcNuL8MsvMP0luKoRtL4a4pLh8xlw55XQ+jwofQT6vAND+zjlEvshYRvUqwCXnAfsg+ohUDMMWp4HGX/D+gbQqi6ccw5IOyh1GBpULkHr8ypR+kZoco6zr9BQeO4ReOAGpwww6kk41AuaecrRoyApCS7wlD+YCBkZUMUzCs/caU6TTYUKTvmr951n7xA/385xPsOSJZ3yj4tO/Ky/n3ti+esPTix/MuXEctsLydG1117Ll19+yf3338+rr77KWWed5crZAOBpD3bpAdQBNp5i2fXAFzg/PC4FVvmyz+bNm+vpWrp0aebrhORUvW/6D1p70PzMR5sXFuvD767R6O9+1592HdKklLTTPlZRkb3OxU1Ghmp6elb54EHVI0ey6rx+vequXc6y1FTVDz9U3bTJKScmqo4erfrjj0750CHV//5Xdd06pxwfrzp4sOratU5561bVu+7KKq9erXrZZaqxsU552jTVWrVUN292yrNnq5Yvr/rLL0550iSnd+bvv53y88875eRkpzxwoGrp0ll1efBB1bPPzir36KFau3ZWuVs31QYNssodO+7Rli2zyg89pNq/f1b56adVX3ghqzxypGp0dFZ51CgnZq+xY1XHjHH+Tqqqb72l+v33WctnzVLdsMF5nZ6uumaN6p49WqiK47/tI0eOaFJSkqo69Vu0aNEJy8+kzsAaPdV39akWnOkD+B9Oe38qzq//nkAfoI9nuQCTgN+An4AWvuy3IBJBRkaGPvj2aq09aL42fm6hvrAgTlf9vl8zMjJOe99FVVH6z7Jzp+qKFc7rjAzV+fNV33sva/kbb6iOGJFVfvJJ1QEDsso9e6r27eu8/uwz1YYNnS9Ir7PPdr5AvXUOD3f2oap6/Ljzr927/7//dsqTJjnlP/5wym++6ZS3bVMtWVL1nXec8saNqvXqqS5e7JTXrVPt2NFJNqpOgmjfXnXHjqzyY49lfTnGxqq+9ppqQoJT/vln1U8/dRKUquru3VlJwxuv5/tAVZ0v2+xJ72RF6XMuLMWtzgsXLtRatWrp008/fcp13EoEbl41dHceyxXo69bxczNz5Q4Wxf1NROlQ5vZrR53KwTWUbX6pOs0X27fDqlXQvj2cfbbTlDFjhtN8UbOm00779tswZ47TnDBxotNEsWePc1o9YoSzjiocPw4ffui0Hd9zj3Oc7793jvHss0752DHnlN0rMhJKl3Ze33ij0wxw/HjW8pEjnTZnrw8+gPM8p/+lSsFPPzltzACVK8ORIxAW5pTPPddpny3huaC6bl3nig2vhg2dJg+vZs1g8eKs8iWXZLVJe8uXZLszpkkT5+FVv77z8Dr7bOfhVaoUJyhht34WWwcOHGDAgAG8/fbbXHTRRVx//fWFH8SpMkRRfZzpGUFSSpq2HPmV1h40X+ev/+u09xUoZs78Xn/+Oav8/feq2c82Bw7MahLYsUP1xhuzfgXv2+f8Sp482SnHxDjlzz93yosWOc0ZK1c65a++Uq1TR/Uvz5916VLVxx/Pal6Ii1P93/+yjv3nn1nrFqTi9kvRF1bnwLR48WKtVq2ahoaG6jPPPJPZLHQqAXdGUFR9uWkPexOO06B6ea5rfHbeGxRBycnOL8YSJWDpUvjmG3juOWfZE0/Ajh3w0UdO+ZVX6lO5Mnz9tVN++GHn12+bNs4VILNnO/vp2tX5Vb9qFVx3nbNuRITTOentRGvdGn74AaKinPLVVzuxeHXqBL//nlXu0IETrnlu0MB5eJ0TXH3YxvxD1apVqVu3LgsXLvTrlW5BlwgWb94LwB0tapxwSWRRcuwYpKc7X8Tx8TB+PNx9t/Ml+s47cO+98OuvzuVmy5c7TSLPPOM0lVSvfmITxYMP/k7z5s0yy9HRTtOM9wqIbduy1hVxLqPzKl0axo7NKoeFQatWrlTZmKCgqrz99tusW7eO8ePH07hxY1asWOH376KgannMUGX1784NKJedX9nP0WSZM8f5YgcnCZQtm3XZ35EjTtv6Rs+ITa1aQb9+Tns5OG3wKSlOEgB48knIPoFb48aHueyyrPLFFztt1UU0BxpTbP3+++907tyZ+++/n9jYWJKSkgD8ngQgyBLBjiMZ7DmSTPXIMOpVLVeox965M2tIhehouPXWrGWvvOJ0mG7eDGXKwH//6zTdANSu7VynffvtTrl+fZgwIau5pmRJ+1I3pihLT09n/PjxNGrUiJUrVzJ58mRiYmIIDw/3d2iZgioR/JHgXILSsk5FV2/0yshwrlBJTHTKr77qfKHv2eOUd+92EsPhw0555kznzkdv+/mQIdCxo/M6JCTrbk5jTOCJj49n6NChXHHFFWzatImHH36YEkXsMrCiFY3L9h1zfpLXdeFyUVXn0kdwvtRbt3ZuPT92zLnU8dlnsy4JfOopp/nG27xTr94/Lxc0xgSu1NRUoqOjycjIoFq1aqxbt47PP/+cWrVq+Tu0HAVVZ/Gh404iqFq+dIHvu1Ur5wzgu++cW/Xnz3eupS9TxvmiHzGiwA9pjCmC1q5dywMPPMCGDRuoXr06nTt35jzvDS1FVFCdEXgTQbWIsDPeV3IydOnifPGD00l7xx3OlT4AV10FjRqd8WGMMQEiKSmJwYMH07p1a/bt28cnn3xC586d/R2WT4LqjCAx1UkEZ5U9vXaYtDRnxMLrrnMurVy/3rl8s21buPPOgozUGBNobr75ZhYtWkSvXr145ZVXqOAdtS4ABFUiOJTsaRqKyH/TkKrzZf/5504nb+nSziWfZW10CmOC1pEjRyhVqhRhYWE8/fTTDBw4kI7eKz0CSFA1DSWnO4mgfFhJn7c5fNgZD0fEGRtn3rysjl1LAsYErwULFtCoUSP++9//AnDFFVcEZBKAIEoEKWkZJKVBSAkhIsy3E6GjR51mnyefdC73DAlxhlWw6/aNCV7x8fF0796d66+/noiICG688UZ/h3TGgqZpKCE5FYDyYaE+3UOwb59z01ZsrDOWTxG96ssYU4i++uorunXrxsGDBxk6dChPP/00pUsX/FWIhS1ozgiSUtMBKFMq79z3wQfOMAzx8c7QDVdd5XZ0xphAUL16dS688ELWrVvH8OHDi0USgCBKBCmemeRLheZd5datnfHnw878KlNjTABTVaZNm0Zfz5yhjRo1Yvny5TRu3NjPkRWsoEkER487ZwThJUNOuc5vvzlDONep40zGbUM7GBO8tm3bRqdOnXjwwQeJi4srUoPEFbSgSQTH05xEULrkqav80EMwbFghBWSMKZLS09MZM2YMjRo1YvXq1UyZMoUlS5YUqUHiClrQdBZnNg2F5JwIVJ1JWLKPx2+MCT7x8fEMHz6cjh078vrrr1OjRg1/h+S6oEkEicfTAChX+p9VTklxhnO+4YasoaKNMcEjJSWFd999l/vuu49q1aoRGxtL7dq1i2UzUE6Cpmko1XMzWU6dxSNHOkM/eydpN8YEj9WrV9O8eXN69uzJ4sWLAahTp07QJAEIokSQluE0DYXm0DR05IgzZWMQfe7GBL1jx47x5JNPcumll3Lw4EHmzp3LNddc4++w/CJomobSPGcEoTncTDZ2rDUJGRNsbrrpJhYvXkzv3r0ZNWoUkd4JQoJQ0JwRZHi+6Uuc9LM/1bnh2M4GjAkChw8fJjk5GYAhQ4bw9ddfM2XKlKBOAhBEicD7i//kL/yGDaFHj8KPxxhTuObPn0/Dhg0ZPnw4AO3bt+fKK6/0c1RFQ9AkgvTMM4Ks99LSnHsH7rjDT0EZY1y3b98+7rnnHm644QYqVqzIrbfe6u+Qipzg6SPI8PQRZOssDg2FJ57wV0TGGLctWrSIbt26cfjwYYYPH87gwYMpZROE/0PQJIIMTyIIydY29MsvcPbZUL68v6Iyxrjp3HPPpUGDBrz++us0bNjQ3+EUWUHTNJSRQ9PQnXeCnSUaU3xkZGQwdepUHn74YQAaNmzIN998Y0kgDz6dEYhICaAJcA6QBGxS1b/dDKygpXvOCLLPRTBjhjPQnDEm8G3dupUHH3yQmJgYrrzySpKSkor1+EAFKddEICLnA4OATsCvwD4gDLhQRI4BU4C3VTXD7UDPlPeqoeyXjzZt6jyMMYErPT2dsWPHMmTIEEqWLMmbb75Jz549g+rO4DOVV9PQSOBd4HxV7ayq/1bVrqp6MXAjEAl0P9XGInKtiGwRka0iMjiH5ZEiMk9E1ovIJhG5/0wqkxvFyQTefxq//ALLljlXDhljAld8fDwjR47k6quvJi4ujl69elkSyKdczwhU9e5clu0Fxp5quYiEAJOAq4FdwGoRmauqcdlW6wvEqeoNIlIF2CIi76lqSj7qkC/efx8ffuiML5SQYPMOGBNojh8/zvz582nfvn3mIHG1atWyBHCafOosFpHbRSTC8/pZEZkjIpfksVkrYKuqbvN8sc8CbjppHQUixPn0ygEHAFd+o588hETPnvDdd5YEjAk0P/zwA82bN+fVV1/NHCQumEYKdYOvl48OUdXZItIO6AyMBl4HWueyzbnAH9nKu3JYfyIwF/gLiADuzKm/QUR6A70BqlWrRkxMjI9hZ9m6zTnJ2LVrFzExezPfP41dBZTExMTT+nsFMqtz8ZSUlMT06dP5+OOPqVy5Ms899xylSpUq9vXOzrXPWVXzfAA/ep5fBO7J/l4u29wOTMtW7g5MOGmdrsAYnKb7C4DfgfK57bd58+Z6OiYt/VVrD5qvLyyI04wM1alTVbdsOa1dBZSlS5f6O4RCZ3Uunjp16qSAPvzww3r48OGgqPPJzqTOwBo9xfeqr/cR/CkiU4A7gAUiUpq8m5V2ATWzlWvg/PLP7n5gjifOrZ5EcJGPMeVL9hvKtm6F3r1hyRI3jmSMKSiHDh3KnCt46NChLFu2jMmTJ1Pe7gItUL4mgjuAL4FrVfUQUBH4Tx7brAbqiUhdESkF3IXTDJTdTqAjgIhUA+oD23yMKV/SPQ1OISWEevXg77+dG8qMMUXT3LlzTxgk7vLLL6d9+/Z+jqp4yjURiEhFEamIc+9ADLDfUz4OrMltW1VNA/rhJJDNwIequklE+ohIH89qI4DLROQnYAkwSFXjz6RCefF2J1WtChUrunkkY8zp2Lt3L3fddRc33XQTlStXpmvXrv4OqdjLq7N4Lc6VPTl1xytwXm4bq+oCYMFJ772R7fVfQKFMCeS9jwBgzhw4cAB69SqMIxtjfLVw4UK6detGYmIiI0aMYNCgQZQsWdLfYRV7ed1HULewAnFb5uWjIsycCTt2WCIwpqipWbMmjRs3ZvLkyURFRfk7nKCR1xATud4roKrrCjYc92TmAeCTTyAx0Z/RGGPAGSRuypQpxMbGMmXKFBo2bBhUl4MWFXk1Db2ayzIFrirAWAqFiPOIiPB3JMYEt19++YVevXqxfPlyrr76apKTkwkLC/N3WEEpr6ahYjePW2qKMxnNnXdCq1b+jsaY4JOWlsarr77Kc889R3h4ODNmzKBHjx52Z7Af+TwxjYg0AqJwriACQFVnuhGUmxIS4Y03oFkzSwTG+MP+/ft5+eWXue6665g0aRLVq1f3d0hBz9f5CJ4DOuAkggVAF+BbIHASgae3uGJFOHrUz7EYE2SOHz9OdHQ0Dz74INWqVWP9+vXUrFkz7w1NofD1hrKuODd+7VHV+3EmqSntWlQukhyvhDXGuGXlypU0a9aMPn368PXXXwNYEihifE0ESeoMBpcmIuWBveRxD0FR471q6M8/4dFHYe/eXFc3xpyhxMREHnvsMdq2bcvRo0dZuHAhnTp18ndYJge+JoI1IlIBeBPnJrN1wCq3gnKD9z6Cfftg5kzIKPJzqhkT2G6++WbGjRtH37592bhxI507d/Z3SOYUfOojUNVHPC/fEJGFOCOEbnAvLPdc0gxmHPR3FMYUTwcPHiQsLIzw8HCGDRvGsGHDaNeunb/DMnnwdWKaW0QkEkBVtwM7ReRmF+NyTXq6nQ0Y44Y5c+YQFRXFsGHDAGjXrp0lgQDha9PQc6p62FvwjED6nCsRucQ71tDyb50+AmNMwdizZw9du3bltttu4+yzz+auu+7yd0gmn3xNBDmt5/M9CEWBt48gMRG++ca/sRhTXHzxxRdERUUxf/58XnjhBVatWkWzZs38HZbJp/x0Fr8mIueLyHkiMgan0zhgeK8auvEGYf16v4ZiTLFRu3ZtmjVrRmxsLE899ZSNFBqgfE0E/YEU4APgQyAJ6OtWUG7I8JwS2F3sxpy+jIwMJk6cyIMPPghAVFQUS5Ys4aKLXJlY0BQSX68aOgoMFpFyqhqY43Z6TgkWLhCS18GAAf4Nx5hAs2XLFnr27Ml3331H586dbZC4YsTXq4YuE5E4IM5TbiIik12NrIB5m4b2/O3MRWCM8U1qaiovvvgiTZo0IS4ujujoaL744gtLAsWIrx2+Y4DOeOYcVtX1IhKQk4f2fAAeusLfURgTOA4ePMgrr7zCDTfcwIQJEzj77LP9HZIpYL72EaCqf5z0VnoBx+IqVc17JWMMAMnJyUyePJmMjAyqVq3Khg0bmD17tiWBYsrXRPCHiFwGqIiUEpEncSakDzgzZsBnn/k7CmOKrm+//ZYmTZrQt2/fzEHiatSo4eeojJt8TQR9cK4SOhfYBTQlwK4a8tq1Cw4d8ncUxhQ9CQkJ9OvXj8svv5yUlBQWLVpkg8QFiTz7CEQkBBirqt0KIR7XPfUU3JjrTMzGBKebb76ZpUuX8uijjzJy5EjKlSvn75BMIckzEahquohUEZFSqppSGEG5qWRJOOssf0dhTNFw4MABwsLCKFOmDCNGjEBEaNOmjb/DMoXM16ah7cB3IjJERAZ4Hy7G5ZopUyA21t9RGON/H330EQ0aNMgcJO6yyy6zJBCkfE0EfwHzPetHZHsEnLQ0f0dgjH/t3r2bW2+9ldtvv52aNWvSrVuxaPU1Z8DXO4uHux1IYenbF5o29XcUxvjH559/zr///W+Sk5N5+eWXGTBgAKGhATV+pHFBrmcEIjJVRBqfYllZEXlAROznhDEB4rzzzqNly5asX7+egQMHWhIwQN5nBJOBIZ5ksBHYB4QB9YDywHTgPVcjLGCTJkGXumBzZ5tgkJ6ezsSJE9mwYQNvvfUWDRo0YNGiRf4OyxQxuSYCVY0F7hCRckALoDrOyKObVXWL++EVvNBQCAnxdxTGuC8uLo5evXqxcuVKrrvuOhskzpySr30EiUCMu6G4yzvCxEMPwTnn+DcWY9yUkpLCqFGjGDFiBBEREbz77rvcc889iI3Bbk7B57GGToeIXCsiW0Rkq4gMPsU6HUQkVkQ2icgyt2LxjjQk2H8GU7wdOnSIMWPGcMsttxAXF0e3bt0sCZhcuZYIPHckTwK6AFHA3SISddI6FXD6IW5U1YbA7W7F4zV6NBw86PZRjClcx48fZ+LEiZmDxP3000/MmjWLqlWr+js0EwDydcmAiJT1TFLji1bAVlXd5tl2FnATnjkNPO4B5qjqTgBV3ZufeE6HiNNPYExx8c0339CrVy927dpFgwYN6NixI+dY+6fJB5++Ej0jj04DygG1RKQJ8JCqPpLLZucC2Yeu3gW0PmmdC4GSIhKDc4PaOFWdmcPxewO9AapVq0ZMTIwvYZ9g1x/HAbj55t9Yu3ZnvrcPVImJiaf19wpkwVLno0ePMnXqVObOnUu1atUYPXo0ISEhQVF3CJ7POTu36uzmxDQ5NUqePClAKNAc6AiEAytF5HtV/eWEjVSnAlMBWrRooR06dPAx7CzLE+Ngx+9ccMH5dLj8vHxvH6hiYmI4nb9XIAuWOl911VXExMTw+OOPc/XVV9OlSxd/h1SoguVzzs6tOvvcSKKqf5zU4ZTXxDS7gOxX69fAGari5HXiPc1NR0XkG6AJ8AsFzHvV0KuvQq/LC3rvxhSO+Ph4ypQpQ5kyZXj++ecRES699NKg+2VsCpabE9OsBuqJSF0RKQXcheeMIpvPgMtFJFREyuA0Hbky4Y16TkaqVHFj78a4S1WZNWsWDRo04LnnngOgTZs2XHrppX6OzBQHZzIxTW79A6hqGtAP+BLny/1DVd0kIn1EpI9nnc3AQmADsAqYpqobT6MePuvRwy6jM4Hlzz//5Oabb+buu++mbt263Hvvvf4OyRQzvjYN1T95YhoRaQt8l9tGqroAWHDSe2+cVH4FeMXHOE6bTVlsAtH8+fPp1q0bqampjB49mscee4wQuzXeFDBfzwgm+PhekTd+nL8jMMZ3F1xwAZdddhkbNmzgiSeesCRgXJHrGYGItAEuA6qcNBFNeSAg/0Wea3NwmyIsPT2d8ePHs379eqKjo7nooov44osv/B2WKebyOiMohXPvQCgnTkhzBOjqbmju6HqbvyMwJmebNm2ibdu2DBgwgPj4eJKTk/0dkgkSeY0+ugxYJiLRqrqjkGIyJqikpKTw0ksvMXLkSCIjI3n//fe56667bHwgU2h87Sw+JiKvAA1x5iMAQFWvciUqF01+HR5o5+8ojMly6NAhxo8fz+23387YsWOpYtc4m0Lma2fxe8DPQF1gOM5k9qtdislVdev4OwJj4NixY4wbN4709PTMQeLee+89SwLGL3xNBJVU9S0gVVWXqeoDQEDeyRJkd+GbImjp0qU0btyYxx57LPOO4OrVq/s3KBPUfE0EqZ7n3SJyvYg0wxkywhjjo8OHD/PQQw9x1VVXISIsXbqUjh07+jssY3xOBCNFJBJ4AngSZyTSx9wKyk1Tp/o7AhOsbr75ZqZNm8Z//vMfNmzYEHQDppmiy9epKud7Xh4GroTMO4sDzvkX+DsCE0z27dtH2bJlKVOmDC+++CIhISG0bNnS32EZc4JczwhEJERE7haRJ0Wkkee9f4nICmBioURYwDoG3HVOJhCpKu+///4Jg8RdeumllgRMkZTXGcFbOENJrwLGi8gOoA0wWFU/dTk2YwLSrl27ePjhh5k/fz6tW7fmvvvu83dIxuQqr0TQArhYVTNEJAyIBy5Q1T3uh+aO6dPtPgLjnrlz5/Lvf/+b9PR0xowZQ//+/W18IFPk5ZUIUlQ1A0BVk0Xkl0BOAgD16/s7AlOcXXjhhbRr146JEydy3nnBMxOeCWx5JYKLRGSD57UA53vKAqiqXuxqdAVIPeNQt7OzAVOA0tLSGDt2LBs2bGDmzJlcdNFFLFiwIO8NjSlC8koEDQolikJko7eYgrJhwwZ69uzJmjVruOmmm0hOTiYsLCzvDY0pYnK9akhVd+T2KKwgC4J3Xpq33/ZrGKYYOH78OM899xzNmzdn586dfPjhh3zyySeWBEzA8vWGsoDnnaEsKsrOCcyZOXLkCJMnT+buu+8mLi6O22+/3UYKNQEtaBKBV+vW/o7ABKKjR48yZswY0tPTqVKlChs3bmTmzJlUqlTJ36EZc8Z8TgQiEi4ids2NCTpLliyhcePGDBgwgGXLlgFQrVo1P0dlTMHxKRGIyA1ALLDQU24qInNdjMs177zj7whMoDh06BC9evWiU6dOhIaGsmzZMq66ym5NN8WPr2cEw4BWwCEAVY0F6rgRkNsaNfJ3BCZQ3HLLLURHRzNo0CDWr19P+/bt/R2SMa7wdYayNFU9XBw6xJo183cEpij7+++/KVeuHGXLluWll14iNDSU5s2b+zssY1zl6xnBRhG5BwgRkXoiMgFY4WJcrsnI8HcEpihSVd555x2ioqIyB4lr3bq1JQETFHxNBP1x5is+DryPMxz1Yy7F5KrZs/0dgSlqdu7cyfXXX8+9995L/fr16dmzp79DMqZQ+do0VF9VnwGecTOYwmB9BCa7zz77jH//+9+oKuPHj+eRRx6xQeJM0PE1EbwmItWB2cAsVd3kYkyuatjQ3xGYokBVEREuuugiOnTowIQJE6hTp46/wzLGL3xqGlLVK4EOwD5gqoj8JCLPuhmYW9LS/B2B8ae0tDRefvllunfvDkD9+vWZN2+eJQET1Hy+oUxV96jqeKAPzj0FQ90Kyk1zA/LuB1MQ1q9fT+vWrRk8eDDHjh0jOTnZ3yEZUyT4ekNZAxEZJiIbcaaoXAHUcDUyl1gfQfBJTk7m2WefpUWLFvz555989NFHzJkzxwaJM8bD1z6CGcD/gGtU9S8X43HdhRf6OwJT2BISEpgyZQrdunXjtddeo2LFiv4OyZgixdc+gktVdVx+k4CIXCsiW0Rkq4gMzmW9liKSLiJd87P/05GS4vYRTFGQmJjI6NGjMweJi4uLIzo62pKAMTnINRGIyIee559EZEO2x0/ZZi471bYhwCSgCxAF3C0iUadY72Xgy9OtRH4sKpSjGH9avXo1jRo1YuDAgXzzzTcAVKlSxc9RGVN05dU09Kjn+V+nse9WwFZV3QYgIrOAm4C4k9brD3wMtDyNY+SbXT5afB04cIAnnniC6Oho6tevz/Lly2nbtq2/wzKmyMs1Eajqbs/LR1R1UPZlIvIyMOifW2U6F/gjW3kXcMJsACJyLnALcBW5JAIR6Q30Bmf435iYmNzCztGffx0HIDXtV2Jitud7+0CVmJh4Wn+vQPToo4+yceNGbr/9dnr16kVqamrQ1D2YPmcvq3PB8bWz+Gr++aXfJYf3sstphDo9qTwWGKSq6bkNaKeqU4GpAC1atNAOHTrkEe4/LT60EXbuoHbtenToUCff2weqmJgYTufvFSj27NlDREQEZcuW5c0336RUqVIcOnSoWNc5J8X9c86J1bng5NVH8LCI/ATUP6mP4Hcg1z4CnDOAmtnKNYCTO5tbALNEZDvQFZgsIjfnpwK+ykh3noPsB0SxpapER0cTFRXF0KHOLS2tWrWiadOm/g3MmACU1xnB+8AXwItA9qt+ElT1QB7brgbqiUhd4E/gLuCe7Cuoal3vaxGJBuar6qc+RZ5P4kl5Uf/orjaBZvv27Tz00EMsWrSIdu3a0bt3b3+HZExAyysRqKpuF5G+Jy8QkYq5JQNVTRORfjhXA4UA01V1k4j08Sx/40wCz68SnpanmjVzX88UbZ988gndu3dHRJg4cSIPP/wwJUoE3dTbxhQoX84I/gWsxWnfz96Qr8B5uW2sqguABSe9l2MCUNX78ojljKindyI5yc2jGLd4B4lr2LAhnTp1Yty4cdSuXdvfYRlTLOT6U0pV/+V5rquq53mevY9ck0BRk5rqPK9Y6d84TP6kpqbywgsv0K1bNwAuvPBCPv30U0sCxhQgX8caaisiZT2v/y0ir4lILXdDK1ghnnOfBg38G4fx3bp162jVqhXPPPMM6enpHD9+3N8hGVMs+dq4+jpwTESaAAOBHcA7rkXlghBPTatX928cJm9JSUk89dRTtGrVij179vDJJ5/wwQcfULp0aX+HZkyx5GsiSFNVxbkzeJyqjgMi3Aur4GV4+giOHfNvHCZvR48e5a233qJHjx7ExcVx8803+zskY4o1XxNBgog8BXQHPveMD1TSvbAKnnfo+bVr/RuHyVlCQgKjRo0iPT2dypUrExcXx1tvvcVZZ53l79CMKfZ8TQR34kxc/4Cq7sEZPuIV16JyQSlP2qpvw1AXOQsXLqRRo0YMHjyY5cuXA1C5cmU/R2VM8PB1GOo9wHtApIj8C0hW1ZmuRlbAQj2dxVWr+TcOk2X//v306NGDLl26ULZsWb777rugGzLAmKLA16uG7gBWAbcDdwA/FMbcAQUpPcN5PnrUv3GYLLfeeivvv/8+Q4YM4ccff6RNmzb+DsmYoOTroHPPAC1VdS+AiFQBFgMfuRVYQTvmSQAbNgCd/BpKUNu9ezcRERGUK1eO0aNHU6pUKZo0aeLvsIwJar72EZTwJgGP/fnYtkgID3eebapK/1BVpk+fToMGDTIHiWvZsqUlAWOKAF/PCBaKyJc48xaD03m8IJf1ixxvH0GlSv6NIxht27aNhx56iMWLF9O+fXv69Onj75CMMdn4lAhU9T8icivQDme8oamq+omrkRWwdM8w1ImJ/o0j2MyZM4fu3bsTEhLC66+/Tu/evW2QOGOKmFwTgYjUA0YD5wM/AU+q6p+FEVhBS/T0EWzahDPNjnGVd5C4xo0bc+211zJ27Fhq2tCvxhRJef00mw7MB27DGYF0gusRuaRMGefZ+gjclZKSwsiRI7nnnntQVerVq8fHH39sScCYIiyvRBChqm+q6hZVHQ3UKYSYXFHSc+5jN6q6Z82aNbRs2ZIhQ4YATlIwxhR9eSWCMBFpJiKXiMglQPhJ5YCRluY8JyT4N47iKCkpiYEDB9K6dWvi4+P57LPP+N///meDxBkTIPLqLN4NvJatvCdbWYGr3AjKDd5O4p9/Bq7xayjFztGjR4mOjqZnz56MGjWKChUq+DskY0w+5JoIVPXKwgrEbWXLOc/WR1Awjhw5wuTJk/nPf/5D5cqV2bx5M5Xs2lxjAlLQXMfn7SOIjPRvHMXB559/TsOGDXnmmWcyB4mzJGBM4AqaRJBqfQRnbN++fXTr1o1//etfREZGsmLFChskzphiIGgSgTcB/Pqrf+MIZLfddhuzZ89m2LBhrFu3jtatW/s7JGNMAfDpzmIREaAbcJ6q/tczX/HZqrrK1egKUIRnPrV69fwbR6D5888/iYyMpFy5cowZM4bSpUvTqFEjf4dljClAvp4RTAbaAHd7ygnAJFcicom3jyAioCbY9B9V5c033yQqKipzkLjmzZtbEjCmGPI1EbRW1b5AMoCqHgRKuRaVC1JSnecjR/wbRyD47bff6NixI71796Z58+b07dvX3yEZY1zkayJI9cxTrJA5H0GGa1G5IMGTALZt828cRd1HH31E48aNWbt2LVOnTmXJkiWcf/75/g7LGOMiXxPBeOAToKqIPA98C7zgWlQuKO+5bLTeBf6No6hSVQCaNGnC9ddfz6ZNm3jwwQdxuoeMMcWZr8NQvycia4GOOMNQ36yqm12NrIB5+wi8N5YZR0pKCi+++CJxcXHMmjWLevXqMXv2bH+HZYwpRL7OWVwLOAbMA+YCRz3vBYzjnvHPrI8gy6pVq2jevDnDhg0jNDTUBokzJkj5OkPZ5zj9AwKEAXWBLUBDl+IqcEcOO8/bt/s1jCLh2LFjDB06lDFjxlC9enXmzZvHv/71L3+HZYzxE1+bhhpnL3tGHn3IlYhc4h1++gLrIyApKYl3332X3r178/LLL1O+fHl/h2SM8aPTurNYVdcBLfNaT0SuFZEtIrJVRAbnsLybiGzwPFaIiGszmXvnLPZOUBNsDh8+zPPPP09aWhqVKlVi8+bNvP7665YEjDE+31k8IFuxBHAJsC+PbUJwbjq7GtgFrBaRuaoal22134ErVPWgiHQBpgKujFuQfNx5PnzYjb0XbfPmzaNPnz7s2bOHtm3b0qFDB86yGXqMMR6+nhFEZHuUxukzuCmPbVoBW1V1m6qmALNO3kZVV3huTgP4Hqjha+D55U0Af/zh1hGKnn379jFixAhuvPFGKlWqxA8//GCDxBlj/iHPMwLPL/tyqvqffO77XCD71+4ucv+13xP44hQx9AZ6A1SrVo2YmJh8hgLHk1M8+/qNmJjt+d4+ED366KPExcVx//33c/fdd5OYmHhaf7tAEyz1zM7qHBzcqnOuiUBEQlU17TSnpczpTiQ9xXGuxEkE7XJarqpTcZqNaNGihZ7Or9olhzbCXzto2PB8OrSpk+/tA8WuXbuoUKEC5cqVY8aMGaxfv57777/f32EVqpiYmKA787E6Bwe36pxX05B3dNFYEZkrIt1F5FbvI49tdwE1s5VrAH+dvJKIXAxMA25S1f2+Bp5fCZ6pKg8fcusI/pWRkcGUKVOIiorKnDz+kksuoW7dun6OzBhT1Pl6H0FFYD/OHMXe+wkUmJPLNquBeiJSF/gTuAu4J/sKnpvS5gDdVfWX/IWeP945i7f97uZR/OPXX3/lwQcfZNmyZXTs2JH+/fv7OyRjTADJKxFU9VwxtJGsBOCVYzNP5kKnSakf8CUQAkxX1U0i0sez/A1gKFAJmOwZ0yZNVVucVk3yUKUysB0aBswtcL6ZPXs29957L6VLl+att97i/vvvt/GBjDH5klciCAHKkY/2/hNWUF0ALDjpvTeyve4F9Mo7zDPnvY+gVEANnn1qqoqI0KxZM2666SZee+01zjnnHH+HZYwJQHklgt2q+t9CicRlx445z4HeR3D8+HGef/55Nm/ezIcffsgFF1zArFmz/B2WMSaA5dVZXGzaGA4dcp73/O3XMM7I999/zyWXXMKIESMIDw+3QeKMMQUir0TQsVCiKATVqjnP553n3zhOx9GjR3n88ce57LLLSEhIYMGCBcycOZPSpUv7OzRjTDGQayJQ1QOFFYjbQkKc55Il/RvH6UhOTmbWrFk88sgjbNq0iS5duvg7JGNMMeLr5aMB7+hR59nbRFTUHTp0iAkTJvDUU09lDhJXoUIFf4dljCmGTmv00UDkTQB79/o1DJ98+umnREVFMXz4cFasWAFgScAY45qgOSM4uzrwR9HuI/j777/p378/s2fPpkmTJsybN4/mzZv7OywTpFJTU9m1axfJycn+DiVHkZGRbN4cUDPmnjFf6hwWFkaNGjUomY928KBJBCGec5/QIlzjrl27smrVKkaOHMnAgQPz9UEaU9B27dpFREQEderUKZI3KSYkJBAREeHvMApVXnVWVfbv38+uXbvyNbxMEf5aLFjesYaKWh/Bzp07Oeuss4iIiGD8+PGULl2aqKgof4dlDMnJyUU2CZiciQiVKlVi375cp4v5h6DpI/DeSBYf79cwMmVkZDBp0iQaNmzI0KFDAWjWrJklAVOkWBIIPKfzmQVNIvCOvlAUBuPcsmULV1xxBf369aNNmzY8+uij/g7JGBPEgiYRlPDU1Hs/gb98+OGHNGnShI0bNzJjxgy+/PJL6tSp49+gjCnCnn/+eRo2bMjFF19M06ZN+eGHH0hLS2P48OHUq1ePpk2b0rRpU55//vnMbUJCQmjatCkNGzakSZMmvPbaa2RkZPixFkVb0PQRHDniPPurj8A7SFzz5s259dZbee211zj77LP9E4wxAWLlypXMnz+fdevWUbp0aeLj40lJSeHZZ59l9+7d/PTTT4SFhZGQkMCrr76auV14eDixsbEA7N27l3vuuYfDhw8zfPhwP9WkaAuaMwLvnMUHCvle6eTkZJ555hm6du2KqnL++efz/vvvWxIwAadDB4iOdl6npjrld991yseOOeUPPnDKhw875TmeGUvi453yvHlOec8e3465e/duKleunDmcSuXKlalQoQJvvvkmr7zyCmFhYQBEREQwbNiwHPdRtWpVpk6dysSJE1HNc9DkoBQ0iaBGDee5MFthVqxYQbNmzXjhhReIiIiwQeKMyadrrrmGP/74gwsvvJBHHnmEZcuWsXXrVmrVqpWvS0fPO+88MjIy2BsId5T6QdA0DXk70ksUQupLTEzk6aefZuLEidSsWZOFCxfSuXNn9w9sjIuyz5lesuSJ5TJlTixHRp5Yrlz5xLKvJ8TlypVj7dq1LF++nKVLl3LnnXfy9NNPn7DOjBkzGDduHPv372fFihXUrFkzx33Z2cCpBc0ZgbdpqDD6CFJSUvjoo4/o27cvGzdutCRgzBkICQmhQ4cODB8+nIkTJzJv3jx27txJQkICAPfffz+xsbFERkaSnp6e4z62bdtGSEgIVatWLczQA0bQJIJDLieCAwcOMGzYMNLS0qhYsSKbN29mwoQJQXfnozEFacuWLfz666+Z5djYWOrXr0/Pnj158sknM4e/SE9PP2XT6759++jTpw/9+vWz+yJOIWiahmrVBP6C2rULft8ff/wxffv2JT4+nquuuor27dsTGRlZ8AcyJsgkJibSv39/Dh06RGhoKBdccAFTp04lMjKSQYMG0ahRIyIiIggPD6dHjx6Z07UmJSXRtGlTUlNTCQ0NpXv37gwYMMDPtSm6giYReH8IFOQPgt27d9OvXz/mzJlDs2bNWLhwIU2bNi24AxgT5Jo3b545Au/Jhg8fzmuvvZbjslM1EZmcBU0i8DYJFWTT0B133MHq1at56aWXeOKJJwgtyiPaGWPMKQTNN5e3j8DbaXy6duzYQcWKFYmIiGDChAmEh4dTv379Mw/QGGP8JGg6i+t4+gZOt48gIyODCRMm0LBhQ4YMGQJA06ZNLQkYYwJe0JwRnImff/6ZXr168d1333Httdfy+OOP+zskY4wpMEFzRuAdWiK/fQSzZs2iSZMmbN68mZkzZ7JgwQJqu3HpkTHG+EnQJALvoHOJib6t7x2psGXLltx+++3ExcXRvXt3uw7ZGFPsBE0i8I4x5B1z6FSSkpIYPHgwt912W+Ygce+++y7VqlVzPUZjzIm8w0k3atSIG264gUMFdNlfdHQ0/fr1K5B9ZdehQwfq16+fOTT2Rx99VODHANi+fTvvv/9+ge0vaBKBL5YvX07Tpk15+eWXqVSpEqmpqf4OyZig5h1OeuPGjVSsWJFJkyb5O6Q8vffee8TGxhIbG0vXrl192iYtLS1fxyjoRBA0ncX79zvP3ikrs0tISGDw4MFMnjyZunXr8tVXX9GpU6dCjc+YoqzO4M9d2e/2l673ed02bdqwYcMGAFatWkX//v1JSUkhPDycGTNmUL9+faKjo5k7dy7Hjh3jt99+45ZbbmHUqFGAMzjdiy++SPXq1bnwwgszh7besWMHDzzwAPv27aNKlSrMmDGDWrVqcd999xEeHs7PP//Mjh07mDFjBm+//TYrV66kdevWRHvH5M7DgQMHeOCBB9i2bRtlypRh6tSpXHzxxQwbNoy//vqL7du3U7lyZcaNG0efPn3YuXMnAGPHjqVt27YsW7YscxZDVeXbb79l8ODBbN68maZNm9KjR48zvoAlaBKBZ3wqjh7757LU1FQ+/fRTHnvsMUaOHEnZsmULNzhjTK7S09NZsmQJPXv2BOCiiy5i4cKFnHXWWSxevJinn36ajz/+GHDGI/rxxx8pXbo09evXp3///oSGhvLcc8+xdu1aIiMjufLKK2nWrBkA/fr1495776VHjx5Mnz6d//u//+PTTz8F4ODBg3z99dfMnTuXG264ge+++45p06bRsmVLYmNjcxxJoFu3boSHhwOwZMkShg0bRrNmzfj000/5+uuvuffeezMnzVm7di3ffvst4eHh3HPPPTz++OO0a9eOnTt30rlzZzZv3szo0aOZNGkSbdu2Zffu3YSHh/PSSy8xevRo5s+fXyB/36BJBHXqwPI9WXMX79+/n3HjxjF06FAqVqzIzz//bAPEGXMK+fnlXpC8YwZt376d5s2bc/XVVwNw+PBhHnnkEX7//XdE5IRm3I4dO2aO9RUVFcWOHTuIj4+nQ4cOVKlSBYA777yTX375BXBmQZvjmUGne/fuDBw4MHNfN9xwAyJC48aNqVatGo0bNwagYcOGbN++PcdE8N5779GiRYvM8rfffpuZpK666ir279/PYc+drTfeeGNm0li8eDFxcXGZ2x05coSEhATatm3LgAED6NatG9dccw3Vq1c/g79ozlztIxCRa0Vki4hsFZHBOSwXERnvWb5BRC5xMx5wTq1mz55NVFQUL774IitXrgSwJGBMEeTtI9ixYwcpKSmZfQRDhgzh8ssvZ+PGjcybNy9zFFIgs8kHnM5mb/u7r1f8ZV/Pu68SJUqcsN8SJUr43K6f0zwI3mNkb33IyMhg5cqVmf0Lf/75JxEREQwePJhp06aRlJREx44d+fnnn306bn64lghEJASYBHQBooC7RSTqpNW6APU8j97A627Fs28fpCXs57Un+3DHHXdQs2ZN1qxZw+WXX+7WIY0xBSQyMpLx48czevRoUlNTOXz4cOZIo7601bdu3ZqYmBj2799Pamoqs2fPzlx22WWXMWvWLMD5Nd+uXbsCjb19+/a89957AMTExFC5cmXKly//j/WuueYaJk6cmFn2Nh/99ttvNG7cmEGDBtGsWbPM1gvvfAwFwc0zglbAVlXdpqopwCzgppPWuQmYqY7vgQoiUvDnPUDiUYj/7GXi1i5j1KhRfP/99zRp0sSNQxljXNCsWTOaNGnCrFmzGDhwIMOGDaNt27Y+jTRavXp1hg0bRps2bejUqROXXJLV+DB+/HhmzJjBxRdfzDvvvMO4ceMKNO5hw4axZs0aLr74YgYPHszbb7+d43rjx4/PXC8qKoo33ngDcDqNGzVqRJMmTQgLC6NLly5cfPHFhIaG0qRJE8aMGXPGMYpb07eJSFfgWlXt5Sl3B1qrar9s68wHXlLVbz3lJcAgVV1z0r5645wxUK1atebe7J0fH25JYdGaLXS5sBy3tax7utUKOImJiZQrV87fYRQqq3PBiIyM5IILLijQfRak9PR0QkJC/B1GofK1zlu3bs3sh/C68sor16pqi5zWd7OzOKcGuZOzji/roKpTgakALVq00A4dOuQ7mA4dIKZ+KU5n20AWExNjdQ4CbtR58+bNRbrvLCEhoUjH5wZf6xwWFpZ5VZQv3Gwa2gVkn0W6BvDXaaxjjDHGRW4mgtVAPRGpKyKlgLuAuSetMxe413P10KXAYVXd7WJMxph8cKvp2LjndD4z15qGVDVNRPoBXwIhwHRV3SQifTzL3wAWANcBW4FjwP1uxWOMyZ+wsDD2799PpUqVbLDFAKGq7N+/n7CwsHxt5+oNZaq6AOfLPvt7b2R7rUBfN2MwxpyeGjVqsGvXLvbt2+fvUHKUnJyc7y+8QOdLncPCwqiR1+iaJwmaO4uNMflTsmRJ6tYtulfYxcTE5KtDtDhwq842+qgxxgQ5SwTGGBPkLBEYY0yQc+3OYreIyD5gx2luXhmIL8BwAoHVOThYnYPDmdS5tqpWyWlBwCWCMyEia051i3VxZXUODlbn4OBWna1pyBhjgpwlAmOMCXLBlgim+jsAP7A6Bwerc3Bwpc5B1UdgjDHmn4LtjMAYY8xJLBEYY0yQK5aJQESuFZEtIrJVRAbnsFxEZLxn+QYRuSSn/QQSH+rczVPXDSKyQkQCfp7OvOqcbb2WIpLumTUvoPlSZxHpICKxIrJJRJYVdowFzYd/25EiMk9E1nvqHNCjGIvIdBHZKyIbT7G84L+/VLVYPXCGvP4NOA8oBawHok5a5zrgC5wZ0i4FfvB33IVQ58uAszyvuwRDnbOt9zXOKLhd/R13IXzOFYA4oJanXNXfcRdCnZ8GXva8rgIcAEr5O/YzqHN74BJg4ymWF/j3V3E8I2gFbFXVbaqaAswCbjppnZuAmer4HqggItULO9AClGedVXWFqh70FL/HmQ0ukPnyOQP0Bz4G9hZmcC7xpc73AHNUdSeAqgZ6vX2pswIR4kyaUA4nEaQVbpgFR1W/wanDqRT491dxTATnAn9kK+/yvJffdQJJfuvTE+cXRSDLs84ici5wC/AGxYMvn/OFwFkiEiMia0Xk3kKLzh2+1Hki0ABnmtufgEdVNaNwwvOLAv/+Ko7zEeQ0ldLJ18j6sk4g8bk+InIlTiJo52pE7vOlzmOBQaqaXkxm2PKlzqFAc6AjEA6sFJHvVfUXt4NziS917gzEAlcB5wNfichyVT3icmz+UuDfX8UxEewCamYr18D5pZDfdQKJT/URkYuBaUAXVd1fSLG5xZc6twBmeZJAZeA6EUlT1U8LJcKC5+u/7XhVPQocFZFvgCZAoCYCX+p8P/CSOg3oW0Xkd+AiYFXhhFjoCvz7qzg2Da0G6olIXREpBdwFzD1pnbnAvZ7e90uBw6q6u7ADLUB51llEagFzgO4B/OswuzzrrKp1VbWOqtYBPgIeCeAkAL792/4MuFxEQkWkDNAa2FzIcRYkX+q8E+cMCBGpBtQHthVqlIWrwL+/it0ZgaqmiUg/4EucKw6mq+omEenjWf4GzhUk1wFbgWM4vygClo91HgpUAiZ7fiGnaQCP3OhjnYsVX+qsqptFZCGwAcgApqlqjpchBgIfP+cRQLSI/ITTbDJIVQN2eGoR+R/QAagsIruA54CS4N73lw0xYYwxQa44Ng0ZY4zJB0sExhgT5CwRGGNMkLNEYIwxQc4SgTHGBDlLBKZI8owWGpvtUSeXdRML4HjRIvK751jrRKTNaexjmohEeV4/fdKyFWcao2c/3r/LRs+ImxXyWL+piFxXEMc2xZddPmqKJBFJVNVyBb1uLvuIBuar6kcicg0wWlUvPoP9nXFMee1XRN4GflHV53NZ/z6ghar2K+hYTPFhZwQmIIhIORFZ4vm1/pOI/GOkURGpLiLfZPvFfLnn/WtEZKVn29kiktcX9DfABZ5tB3j2tVFEHvO8V1ZEPveMf79RRO70vB8jIi1E5CUg3BPHe55liZ7nD7L/QvecidwmIiEi8oqIrBZnjPmHfPizrMQz2JiItBJnnokfPc/1PXfi/he40xPLnZ7Yp3uO82NOf0cThPw99rY97JHTA0jHGUgsFvgE5y748p5llXHuqvSe0SZ6np8AnvG8DgEiPOt+A5T1vD8IGJrD8aLxzFcA3A78gDN4209AWZzhjTcBzYDbgDezbRvpeY7B+fWdGVO2dbwx3gK87XldCmcUyXCgN/Cs5/3SwBqgbg5xJmar32zgWk+5PBDqed0J+Njz+j5gYrbtXwD+7XldAWcMorL+/rzt4d9HsRtiwhQbSara1FsQkZLACyLSHmfohHOBasCebNusBqZ71v1UVWNF5AogCvjOM7RGKZxf0jl5RUSeBfbhjNDaEfhEnQHcEJE5wOXAQmC0iLyM05y0PB/1+gIYLyKlgWuBb1Q1ydMcdbFkzaIWCdQDfj9p+3ARiQXqAGuBr7Kt/7aI1MMZibLkKY5/DXCjiDzpKYcBtQjs8YjMGbJEYAJFN5zZp5qraqqIbMf5Esukqt94EsX1wDsi8gpwEPhKVe/24Rj/UdWPvAUR6ZTTSqr6i4g0xxnv5UURWaSq//WlEqqaLCIxOEMn3wn8z3s4oL+qfpnHLpJUtamIRALzgb7AeJzxdpaq6i2ejvWYU2wvwG2qusWXeE1wsD4CEygigb2eJHAlUPvkFUSktmedN4G3cKb7+x5oKyLeNv8yInKhj8f8BrjZs01ZnGad5SJyDnBMVd8FRnuOc7JUz5lJTmbhDBR2Oc5ganieH/ZuIyIXeo6ZI1U9DPwf8KRnm0jgT8/i+7KtmoDTROb1JdBfPKdHItLsVMcwwcMSgQkU7wEtRGQNztnBzzms0wGIFZEfcdrxx6nqPpwvxv+JyAacxHCRLwdU1XU4fQercPoMpqnqj0BjYJWnieYZYGQOm08FNng7i0+yCGde2sXqTL8IzjwRccA6cSYtn0IeZ+yeWNbjDM08Cufs5Duc/gOvpUCUt7MY58yhpCe2jZ6yCXJ2+agxxgQ5OyMwxpggZ4nAGGOCnCUCY4wJcpYIjDEmyFkiMMaYIGeJwBhjgpwlAmOMCXL/D+76zW0FyBz9AAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(FPR, TPR, 'b:', label='SGD')\n",
    "plot_roc_curve(FPR_forest, TPR_forest, 'Random Forest')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> as you can see RandomForest's ROC curve looks better than SGD's : it come closer to the top-left corner"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9983400347444625"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "roc_auc_score(y_train_5, y_scores_forest)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([1, 0, 0, ..., 1, 0, 0]),\n",
       " array([0.89, 0.01, 0.04, ..., 0.98, 0.08, 0.06]))"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train_5, y_scores_forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Pecision forest: 0.9905083315756169\n",
      "Recall forest: 0.8662608374838591\n"
     ]
    }
   ],
   "source": [
    "print('Pecision forest:',precision_score(y_train_5,y_scores_forest.round()))\n",
    "print('Recall forest:',recall_score(y_train_5,y_scores_forest.round()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> 99% precision and 86% recall , Not too bad!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Multiclass Classification\n",
    "\n",
    "Logistic Regrssion cl, Random Forest cl, naive Bayes cl are capable of handling multiple classes natively.\n",
    "\n",
    "have 10 classes of image to train 10 binart classifier and select the highest score. this is called the *one-versus-the-res (OvR)* strategy (also called one-versus-all)\n",
    "\n",
    "another stategy is to train binary classifier for every pair of digits:0s and 1s, 0s and 2s, 1s and 2s, and so on this is called *one-versus-one (OvO)* strategy. you wlii train N*(N-1)/2 times if have N classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([9], dtype=uint8)"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.svm import SVC\n",
    "\n",
    "svm_clf = SVC()\n",
    "svm_clf.fit(x_train,y_train)\n",
    "svm_clf.predict([x_test[12]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "9"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test[12]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> under the hood, Scikit-Learn actually used the OvO strategy: it train 45 binary classifiers, got their decision sscores for the image, and selected the class that won the most duels."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 1.70901642, -0.30875332,  2.72613485,  5.12738826,  8.29913769,\n",
       "         4.80115581,  0.69848667,  7.28801973,  5.21585844,  9.3152483 ]])"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "some_digit_scores = svm_clf.decision_function([x_test[12]])\n",
    "some_digit_scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(9, array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9], dtype=uint8))"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.argmax(some_digit_scores), svm_clf.classes_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(x_test[12].reshape(28,28));"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "if you want to force to use OvR or OvO it's simply\n",
    "* OneVsRestClassifier\n",
    "* OneVsOneClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OneVsRestClassifier(estimator=SVC())"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.multiclass import OneVsRestClassifier\n",
    "\n",
    "ovr_clf = OneVsRestClassifier(SVC())\n",
    "ovr_clf.fit(x_train,y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([9], dtype=uint8)"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ovr_clf.predict([x_test[12]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-2.60820475, -3.09580555, -2.65879979, -1.55235189, -1.53312881,\n",
       "        -2.19611651, -3.3470251 , -1.93431147, -2.02584848,  1.48961706]])"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "some_digit_ovr = ovr_clf.decision_function([x_test[12]])\n",
    "some_digit_ovr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(9, array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9], dtype=uint8))"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.argmax(some_digit_ovr), ovr_clf.classes_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Training an SGDClassifier is just as easy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([9], dtype=uint8)"
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sgd_clf.fit(x_train, y_train)\n",
    "sgd_clf.predict([x_test[12]])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> this time Scikit-Learn used the OvR strategy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ -8.12730898, -11.88107109,  -6.46080517,  -4.39046016,\n",
       "         -2.96234373,  -3.77821801,  -7.64083916,  -2.49099509,\n",
       "         -5.11510566,   1.48092261]])"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sgd_clf.decision_function([x_test[12]])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> you can see that the classifier is fairly confident about its prediciton: all score are negative except 9s but can assume the 7s is close to 9s becuse score are is top2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.90905, 0.90645, 0.91525])"
      ]
     },
     "execution_count": 71,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cross_val_score(sgd_clf, x_train, y_train, cv=3, scoring='accuracy')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> accuracy above 90% but how can we improve? . using *StrandardScaler*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The idea behind StandardScaler is that it will transform your data such that its distribution will have a mean value 0 and standard deviation of 1."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.04313725, 0.74509804, 0.99215686,\n",
       "       0.2745098 , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.1372549 , 0.94509804, 0.88235294, 0.62745098])"
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train[0][340:380]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([-0.1273685 , -0.22614336, -0.35220291, -0.50758351, -0.67944964,\n",
       "       -0.82825105, -0.88883004, -0.74022511,  1.06068287,  1.68492907,\n",
       "       -0.17757569, -0.91768349, -1.02909594, -1.058279  , -0.96567404,\n",
       "       -0.79447676, -0.61674936, -0.47007288, -0.35512441, -0.25540754,\n",
       "       -0.15382451, -0.05245428, -0.02150587, -0.00992198, -0.00408252,\n",
       "       -0.01093439, -0.0255448 , -0.05707308, -0.12466202, -0.23849817,\n",
       "       -0.3732589 , -0.53313607, -0.70209414, -0.83745005, -0.88007425,\n",
       "       -0.84101427, -0.48769493,  1.32485786,  0.99837547,  0.32281427])"
      ]
     },
     "execution_count": 90,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "scaler = StandardScaler()\n",
    "x_train_scaled = scaler.fit_transform(x_train.astype(np.float64))\n",
    "x_train_scaled[0][340:380]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.8983, 0.891 , 0.9018])"
      ]
     },
     "execution_count": 91,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cross_val_score(sgd_clf, x_train_scaled, y_train, cv=3, scoring='accuracy')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> ahh yeah sometimes \n",
    "\n",
    "> I know this will increase the accuracy, if the data is not normallize yet."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Error Analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[5697,    2,   32,   20,   18,   59,   39,   15,   38,    3],\n",
       "       [   2, 6506,   50,   29,    7,   39,   10,   15,   70,   14],\n",
       "       [  31,   43, 5350,  103,   83,   38,   60,   86,  140,   24],\n",
       "       [  21,   27,  206, 5405,   10,  200,   28,   65,   97,   72],\n",
       "       [  13,   17,   28,   17, 5436,   14,   48,   15,   41,  213],\n",
       "       [  48,   17,   53,  206,   80, 4740,   89,   24,   94,   70],\n",
       "       [  38,   14,   48,    5,   40,  125, 5619,    4,   22,    3],\n",
       "       [  18,   19,   69,   17,   87,   13,    3, 5844,   13,  182],\n",
       "       [  47,  110,  114,  216,   66,  263,   57,   36, 4808,  134],\n",
       "       [  30,   15,   23,  114,  219,   61,    1,  227,   49, 5210]],\n",
       "      dtype=int64)"
      ]
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train_pred = cross_val_predict(sgd_clf, x_train, y_train, cv=3)\n",
    "conf_mx = confusion_matrix(y_train, y_train_pred)\n",
    "conf_mx"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 142,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1080x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def plot_conf_mx(conf_mx,fmt):\n",
    "    cm_df = pd.DataFrame(conf_mx,columns=sgd_clf.classes_,index=sgd_clf.classes_)\n",
    "    plt.figure(figsize=(15,10))\n",
    "    sn.heatmap(cm_df,annot=True,fmt=fmt);\n",
    "    plt.ylabel('True')\n",
    "    plt.xlabel('Predict')\n",
    "    plt.show()\n",
    "\n",
    "plot_conf_mx(conf_mx,'d')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "percentage confusion matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([5923, 6742, 5958, 6131, 5842, 5421, 5918, 6265, 5851, 5949],\n",
       "      dtype=int64)"
      ]
     },
     "execution_count": 122,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conf_mx.sum(axis=1, keepdims=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[5923],\n",
       "       [6742],\n",
       "       [5958],\n",
       "       [6131],\n",
       "       [5842],\n",
       "       [5421],\n",
       "       [5918],\n",
       "       [6265],\n",
       "       [5851],\n",
       "       [5949]], dtype=int64)"
      ]
     },
     "execution_count": 123,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "row_sums = conf_mx.sum(axis=1, keepdims=1)\n",
    "row_sums"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 149,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1080x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "percent_cm = (conf_mx / row_sums)*100\n",
    "\n",
    "plot_conf_mx(percent_cm,fmt='.2f')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Until now each instance has always been assigned to just one class. Consider a face-recognition: it shold attach tag for person in the same photo such as : Alice, bob, charlie will return output [1,0,1] mean no Bob "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0, 1],\n",
       "       [0, 0],\n",
       "       [0, 0],\n",
       "       ...,\n",
       "       [0, 1],\n",
       "       [0, 0],\n",
       "       [1, 0]])"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "\n",
    "y_train_large = (y_train >= 7)*1\n",
    "y_train_odd = (y_train %2 == 1)*1\n",
    "y_multilabel = np.c_[y_train_large, y_train_odd]\n",
    "y_multilabel"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> numpy. c_ = <numpy.lib.index_tricks.CClass object> Translates slice objects to concatenation along the second axis. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "using KNeighborsClassifier which support multilabel classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "KNeighborsClassifier()"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "knn_clf = KNeighborsClassifier()\n",
    "knn_clf.fit(x_train, y_multilabel)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0, 1]])"
      ]
     },
     "execution_count": 159,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "knn_clf.predict([x_train[12]]) # < 7 and odd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3"
      ]
     },
     "execution_count": 161,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train[12]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "there are many ways to evaluate a multilabel classifier, and selecting the right metric depends on your project. One approach is to measure the F1 score, this will compute the average scire for all labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0, 1],\n",
       "       [0, 0],\n",
       "       [0, 0],\n",
       "       ...,\n",
       "       [0, 1],\n",
       "       [0, 0],\n",
       "       [1, 0]])"
      ]
     },
     "execution_count": 162,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train_knn_pred = cross_val_predict(knn_clf, x_train, y_multilabel)\n",
    "y_train_knn_pred"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Macro F1-score (short for macro-averaged F1 score) is used to assess the quality of problems with multiple binary labels or multiple classes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.977511478985599"
      ]
     },
     "execution_count": 167,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "f1_score(y_multilabel, y_train_knn_pred, average='macro')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> this assume that all labels are equally important. In particular if you have many more picture of Alice than Bob or Charlie, you may want to give more weight to the score on pictures of Alice to do this use *average = 'weighted'*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Multioutput Classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "let's build a system that removes noise from images. Notice that the classifier's output is multilabel(1 label per pixel)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "the data original work, but normalize data won't work"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "X, y = mnist['data'], mnist['target']\n",
    "\n",
    "X = np.array(X)\n",
    "y = np.array(y, dtype=np.uint8)  #dtype=np.uint8 change to integer\n",
    "x_train, x_test, y_train, y_test = X[:60000], X[60000:], y[:60000], y[60000:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "noise = np.random.randint(0,100, (len(x_train), 784)) \n",
    "x_train_mod = x_train + noise\n",
    "noise = np.random.randint(0,100, (len(x_test), 784)) \n",
    "x_test_mod = x_test + noise\n",
    "\n",
    "y_train_mod = x_train\n",
    "y_test_mod = x_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x1ed86f12cb0>"
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(y_train_mod[45].reshape(28,28),cmap='binary')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x1ed86f7dfc0>"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD4CAYAAAAq5pAIAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAAsTAAALEwEAmpwYAAAXnklEQVR4nO2da4zdZbXGn9XLtGVaeqf3Mi0UerM3p02xhECQCpiIJHhCE00xeOoHTTRqPMQTlagfkBw0TTwYyxGtRlATJGLAA7RWoFaBoRl6G9pOL06n0/uFzpROp51Z58PsJiPM+6xhLntPzvv8ksnM7GfW3u/+7/3Mf++93rWWuTuEEP//GVDqBQghioPMLkQmyOxCZILMLkQmyOxCZMKgYt7Y6NGjffLkyUl98ODBNP7ixYtJLcoqDBw4kOqXL1+m+tChQ6nOGDSIH+bGxkaqjxgxotu3zY4ZEB+XtrY2qre0tFC9vLyc6ozW1laqR2tj983MaGx03MrKyqgerY3ply5dorHsfjU0NODs2bOd3rkemd3M7gSwFsBAAP/j7o+wv588eTJ+97vfJfVrrrmG3t6BAweSWvTgjBo1iuonTpyg+rx585Ja9I9mzJgxVH/llVeoftttt1GdPXEOHjxIY6N/JM3NzVRnjwkA3HTTTUktMkT0T/DChQtUHzlyZFKL/gHX1tZSvaKigurvvfce1dl9O3r0KI1lz+XPfvazSa3bL+PNbCCA/wZwF4C5AFaZ2dzuXp8Qom/pyXv2ZQBq3X2/u7cA+C2Ae3pnWUKI3qYnZp8C4FCH3+sLl/0LZrbGzKrMrOrMmTM9uDkhRE/oidk7+xDgA29e3X2du1e6e+Xo0aN7cHNCiJ7QE7PXA5jW4fepABp6thwhRF/RE7O/CWCWmc0wszIA9wN4rneWJYTobbqdenP3y2b2ZQAvoj319qS772Qxra2tOHfuXFKfPXs2vc3hw4cntf3799PY+fPnU72pqYnqW7dupTpjwYIFVF+yZAnVoxTV2bNnk9r58+dpbJQnf/fdd6nOUmsAMGzYsKQWpafY4w3EeyOOHTuW1KJ06IQJE6heU1ND9RkzZlB9wID0eXb8+PE0dubMmUmN7QfpUZ7d3V8A8EJPrkMIURy0XVaITJDZhcgEmV2ITJDZhcgEmV2ITJDZhciEotazl5eXo7KyMqlv2bKFxi9fvjypsbwlEJfA7tixg+qs/Lauro7GRnXZY8eOpXpUAsvy+FEJa1Q7PWXKB8od/oWTJ09SndV9R+WxLJ8MxLXyrFR04sSJNJbtD+jKbUfluez2o8ds8+bNSY3tq9CZXYhMkNmFyASZXYhMkNmFyASZXYhMkNmFyISipt5aWlpw6NChpB615z19+nS3NCDu2Ll06VKqs26kUTtmVtYLANu2baN61GK7oSHdM+SGG26gsfv27aP6kCFDqL5nzx6qs2MTpf16Wn7L2LmTVmOHqbeo6xJ7TAD+mEZpYNaOnXlIZ3YhMkFmFyITZHYhMkFmFyITZHYhMkFmFyITZHYhMqGoefaysjJce+21ST1qmczKTKN8cDS1MxrJfPjw4aQWlSRGewCiiaBTp06lOss3RxNmo1x1tEfglltuofqrr76a1KJ9FdG4sChPz447ex4CvA01AMydy2eYRiOhp02bltSikme2B0B5diGEzC5ELsjsQmSCzC5EJsjsQmSCzC5EJsjsQmRCUfPsZkbrm6MRvW+99VZSu+6662hsbW0t1aO86dVXX53U9u7dS2NnzZpF9ag2uqqqiupsj0GU44/GRUePyYkTJ6jO9i9EI5ejcdPR3gk2pjt6PixcuJDqUY6f1ZxHRP0NujsGu0dmN7ODABoBtAK47O7ppvBCiJLSG2f229ydTwoQQpQcvWcXIhN6anYH8JKZvWVmazr7AzNbY2ZVZlYVvb8TQvQdPTX7CndfAuAuAF8ysw9URbj7OnevdPfK8ePH9/DmhBDdpUdmd/eGwvfjAJ4FsKw3FiWE6H26bXYzKzezEVd+BrASAO+BK4QoGT35NH4CgGcLdbuDADzl7v/LAi5cuEB7Yke11Wx8cDQWefHixVSPenWznG207ihfXF1dTfWoZvzUqVNJ7cKFCzQ26s3O6vgBoKamhuoTJkxIatH+A3a/AGDXrl1U/+hHP5rUmpubaWw0Tjra1xEdt/r6+qR244030lj2fGP7WLptdnffD4DvPBBC9BuUehMiE2R2ITJBZhciE2R2ITJBZhciE4pa4hoxatQoqrM0zttvv01jo7HKkyZNovqGDRuSGit/BYCDBw9SPYpnpb0AH8vMUoYA8IMf/IDqf/rTn6gepZgmTpyY1KJW0Z///OepHrVrZimq++67j8ZGqdio9Hf37t1UZ2nBqHSXwZ7nOrMLkQkyuxCZILMLkQkyuxCZILMLkQkyuxCZILMLkQlFzbMPGTIEM2bMSOpNTU00nuWbL168SGO3b99O9auuuorqrO1xlMNn43mj6+6K/vzzzye1X/ziFzQ2Om7R6GLW1hgAZs+e3e3r/tnPfkZ1lsMHgH/84x9JbeXKlTT2yJEjVI9KXD/ykY9Q/ejRo0ktul/vvPNOUmOluzqzC5EJMrsQmSCzC5EJMrsQmSCzC5EJMrsQmSCzC5EJRc2zDxgwgOazo/a9K1asSGosbwkA11xzDdU3bdpEdVa/zGqTAaCuro7q0dofffRRqh86dCipRbXRbMQvAKxfv57qgwcPpjrbV/HnP/+Zxn7nO9+hOhtVDfC9E1Eevby8nOpbt26l+tKlS6nORlm/+eabNJbdrwED0udvndmFyASZXYhMkNmFyASZXYhMkNmFyASZXYhMkNmFyISi5tnPnTuHv/zlL0l9yZIlNP61115LalEP8ah2et68eVRvaGhIatG46LFjx1J99erVVI/GIldUVCQ1Vk8OAKtWraL6TTfdRHV3p/revXuT2ubNm2lsNNKZ5fAB4Bvf+EZSi+r4ozHb0XGNehyMGTMmqUVzBBYuTA9PZvsDwjO7mT1pZsfNbEeHy8aY2ctmtrfwfXR0PUKI0tKVl/G/BHDn+y57CMBGd58FYGPhdyFEPyY0u7u/CuD0+y6+B8CVfZTrAXy6d5clhOhtuvsB3QR3PwIAhe/JjedmtsbMqsysis3eEkL0LX3+aby7r3P3SnevHDlyZF/fnBAiQXfNfszMJgFA4fvx3luSEKIv6K7ZnwNwJV+0GsAfe2c5Qoi+Isyzm9nTAG4FMM7M6gF8F8AjAH5vZg8CqAPwma7c2LBhwzB37tykznpeA7xGeN++fTQ2ql+O8qas73x9fT2N/clPfkL1v//971SPaqtZPnnOnDk0durUqVQ/deoU1aM+AY8//nhSe/bZZ2ksmzsPAN/85jepznq7nzhxoke3He2diGrSWb99lkcHgMbGxqTW1taW1EKzu3tq18XtUawQov+g7bJCZILMLkQmyOxCZILMLkQmyOxCZEJRS1wHDRqE8ePHJ/WonTMreYxSQFGqZMuWLVRnI3jZKGkAeOKJJ6gelciuXbuW6h//+MeTWtTqOUr7sVJMAHjoIV4DxVKe48aNo7G7du2iem1tLdVZ6o2VLANxmenp0+8vF/lw8awN9saNG2ksSxOz8d46swuRCTK7EJkgswuRCTK7EJkgswuRCTK7EJkgswuRCUXNs7s7LWNduXIljb906VJSi0YTHz58mOpR61/Wzjlqp8zWDcTllGxcNABUVVUltQULFtDYaJw0K1EFgD179lC9qampWxoAPPDAA1SPSn9ZGWv0mLFSUQCYP38+1aP236x0eObMmTSWjXvWyGYhhMwuRC7I7EJkgswuRCbI7EJkgswuRCbI7EJkQlHz7JcuXcLJkyeTejRW+ejRo0ktyntGefQoZ3vu3LmkFtWzsxp+IK61j2qvy8rKktoPf/hDGhvtT2DXDQDr16/vts7GOQNxLf7tt/MGx9XV1Uktmk4U5dmjUWasZToAHDhwIKlF46RHjBiR1NjzXGd2ITJBZhciE2R2ITJBZhciE2R2ITJBZhciE2R2ITKhqHn2AQMG0LxtS0sLja+oqEhqUR49yotOnz6d6mfOnElqt912G4194403qM72DwDA888/3+34d955h8Y++OCDVL/vvvuoHh03dtyj3uvf/va3qR7tb1iyZElSO3jwII2NRngfOnSI6v/85z+pzo5bNAOB7flg+wPCM7uZPWlmx81sR4fLHjazw2ZWXfi6O7oeIURp6crL+F8CuLOTy3/s7osKXy/07rKEEL1NaHZ3fxUAf70lhOj39OQDui+b2bbCy/zRqT8yszVmVmVmVazvlhCib+mu2X8K4DoAiwAcAfBY6g/dfZ27V7p7ZVTwIYToO7pldnc/5u6t7t4G4AkAy3p3WUKI3qZbZjezSR1+vRfAjtTfCiH6B2Ge3cyeBnArgHFmVg/guwBuNbNFABzAQQBf7MqNXbx4Efv27Uvqy5bxFwgsl87mUgO81zbA+3gDfJ72VVddRWM/+clPUv373/9+t28bAC5cuJDU7rrrLhr7qU99iupRbfWLL75I9U2bNiW1pUuX0tio1j7Khb/++utJ7frrr6exUa9/NnceiNfOHtO//vWvNHbhwoVJjfXDD83u7qs6ufjnUZwQon+h7bJCZILMLkQmyOxCZILMLkQmyOxCZEJRS1yHDBlCUx5ReowRlRxG1x3Fs91/U6dOpbHDhg2j+v3330/1KP11/PjxpBa1Y45STFGp5mOPJTdPAuDjh0ePTu6yBgBMnDiR6qzUE+CjsHft2kVj6+vrqR6tPXq+sVLURYsW0dgozZxCZ3YhMkFmFyITZHYhMkFmFyITZHYhMkFmFyITZHYhMqGoefaI8+fPU/21115LagMG8P9bK1eupHo0LnrMmDFJjeW5gThPHrXB3rlzJ9VZPrq5uZnGsnbLALBhwwaqR9ff2NiY1B5//HEau3v3bqpHuW7Wtvzmm2+msVGr6aisOWqTfeLEiaQWPR/Y/WL7JnRmFyITZHYhMkFmFyITZHYhMkFmFyITZHYhMkFmFyITippnb2tro22Po9pplkuP2hKz+mEAmDZtGtXr6uqSWpTvjUbwVlVVUX3x4sVUZ3n4+fPn09izZ89S/ZVXXqF6VMvP9jewfDEAVFZWUj1aO+sjEI17jqYXjRw5kuqHDx+melNTU1KLnk8sx888ojO7EJkgswuRCTK7EJkgswuRCTK7EJkgswuRCTK7EJlQ9L7xFRUVSX3KlCk0vqWlJalF9cPV1dVUZ/3No+sfMWIEjY3G+0Z5eFb7DPDjFtVdR2t76aWXqB6t/ROf+ERSi8Ymb9++nerDhw+nOuuPsHz5chprZlSPegxEvd1ZD4JZs2bRWNYjgNXCh2d2M5tmZpvMrMbMdprZVwqXjzGzl81sb+E73wkghCgpXXkZfxnA1919DoDlAL5kZnMBPARgo7vPArCx8LsQop8Smt3dj7j71sLPjQBqAEwBcA+A9YU/Ww/g0320RiFEL/ChPqAzswoAiwG8DmCCux8B2v8hAOj0zZuZrTGzKjOrit57CiH6ji6b3cyGA3gGwFfdnU/U64C7r3P3SnevHD9+fHfWKIToBbpkdjMbjHaj/8bd/1C4+JiZTSrokwDwFqtCiJISpt6sPQfxcwA17v6jDtJzAFYDeKTw/Y/RdbW1tdG2ylHZIUsjTZ8+ncbOmzeP6u+99x7VWcvlhoYGGrtgwQKq79ixg+pRGmfSpElJbf/+/TT2qaeeonrU1vhjH/sY1VetWpXUotbhUWouatHN0mPRKOurr76a6tG46Oi+sdQba5kOAHfccUdSGzQobemu5NlXAPgcgO1mVl247FtoN/nvzexBAHUAPtOF6xJClIjQ7O6+GUBqh8HtvbscIURfoe2yQmSCzC5EJsjsQmSCzC5EJsjsQmRCUUtcW1pa6CjcaIcdyzdHrX//9re/UX3cuHFUZy16oxLXM2fOUL0nY4+j649KWJ9++mmqs7wtAHzta1+jOmsPHpU0R2XJs2fPpjorp47aNR89epTqrNwaAG699Vaqsz0AUXtuti+DtWrXmV2ITJDZhcgEmV2ITJDZhcgEmV2ITJDZhcgEmV2ITCh6np2NsmV12QAwdOjQpBbVNketgaP6ZNZS69KlSzS2p+Oio9HELP573/sejY3quqN2zaNGjaI624OwYcMGGhvV0ke5brbHIKrzHzJkCNWj+11bW0t1Nk468gHLs7e2tiY1ndmFyASZXYhMkNmFyASZXYhMkNmFyASZXYhMkNmFyISi5tkHDhxI867z58+n8QcOHEhqUS56woQJVD906BDVWT377bfzJrt1dXVUj2qno7rvL3zhC0mN1ZMD8f1eu3Yt1aNafnbcWe90AHj33XepHs0ZYOOko7HI0b4LtucDiOvlt2zZktSi/QOsxwDbT6IzuxCZILMLkQkyuxCZILMLkQkyuxCZILMLkQkyuxCZ0JX57NMA/ArARABtANa5+1ozexjAvwO4Uuj9LXd/gV3X0KFDMWfOnKR++vRpuhZWAxzl2WfMmEH1KJfN6vCbmppobNQ3PupZ/8wzz1B9165dSS3qxX/vvfdSneXwAWDPnj1UZ/nmKE8e5fCjmnNWqx/lwaP57Ky/AQCUlZVR/YYbbkhqrF4d4Peb9U7oyqaaywC+7u5bzWwEgLfM7OWC9mN3/68uXIcQosR0ZT77EQBHCj83mlkNAH4aFEL0Oz7Ue3YzqwCwGMDrhYu+bGbbzOxJM+v0dZGZrTGzKjOrOnXqVM9WK4ToNl02u5kNB/AMgK+6+zkAPwVwHYBFaD/zP9ZZnLuvc/dKd6+M5rEJIfqOLpndzAaj3ei/cfc/AIC7H3P3VndvA/AEgGV9t0whRE8JzW7tZTQ/B1Dj7j/qcHnHFpj3AuAfIQohSkpXPo1fAeBzALabWXXhsm8BWGVmiwA4gIMAvhhdUXNzM2pqapJ69DKfpZFmzpxJY1l6CojTY6ykcfr06TQ2KuWM2jlH923y5Mndjl2/fj3Vz58/T/Ubb7yR6iydykYqA7xEFYjLc1mKKvr8KEqdRbfNRjIDwPXXX5/UorQgG1VdXl6e1LryafxmAJ0VydKcuhCif6EddEJkgswuRCbI7EJkgswuRCbI7EJkgswuRCYUtZV0a2srLQeN2vOynHFUFhjlmxsaGqjORhe/8cYbNJblVIE45xuVqW7dujWpNTc309jjx49TnbXv7gpsj0F9fT2NbWxspHrUcpmVJS9dupTGstHHQNyaPGqDzVqTR9fNHlNW4qozuxCZILMLkQkyuxCZILMLkQkyuxCZILMLkQkyuxCZYO5evBszOwGg4wzhcQBOFm0BH47+urb+ui5Aa+suvbm2a929040ZRTX7B27crMrdK0u2AEJ/XVt/XRegtXWXYq1NL+OFyASZXYhMKLXZ15X49hn9dW39dV2A1tZdirK2kr5nF0IUj1Kf2YUQRUJmFyITSmJ2M7vTzHabWa2ZPVSKNaQws4Nmtt3Mqs2sqsRredLMjpvZjg6XjTGzl81sb+E7bzJe3LU9bGaHC8eu2szuLtHappnZJjOrMbOdZvaVwuUlPXZkXUU5bkV/z25mAwHsAXAHgHoAbwJY5e58ikORMLODACrdveQbMMzsFgBNAH7l7vMLlz0K4LS7P1L4Rzna3f+jn6ztYQBNpR7jXZhWNKnjmHEAnwbwAEp47Mi6/g1FOG6lOLMvA1Dr7vvdvQXAbwHcU4J19Hvc/VUA7x+pcg+AK2Nc1qP9yVJ0EmvrF7j7EXffWvi5EcCVMeMlPXZkXUWhFGafAqDj7Jx69K957w7gJTN7y8zWlHoxnTDB3Y8A7U8eAHxGUvEJx3gXk/eNGe83x6474897SinM3tkoqf6U/1vh7ksA3AXgS4WXq6JrdGmMd7HoZMx4v6C74897SinMXg9gWoffpwLg3R6LiLs3FL4fB/As+t8o6mNXJugWvvOOkUWkP43x7mzMOPrBsSvl+PNSmP1NALPMbIaZlQG4H8BzJVjHBzCz8sIHJzCzcgAr0f9GUT8HYHXh59UA/ljCtfwL/WWMd2rMOEp87Eo+/tzdi/4F4G60fyK/D8B/lmINiXXNBPB24WtnqdcG4Gm0v6y7hPZXRA8CGAtgI4C9he9j+tHafg1gO4BtaDfWpBKt7Wa0vzXcBqC68HV3qY8dWVdRjpu2ywqRCdpBJ0QmyOxCZILMLkQmyOxCZILMLkQmyOxCZILMLkQm/B+lpCinpdHdLAAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(x_train_mod[45].reshape(28,28),cmap='binary')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "knn_clf.fit(x_train_mod,y_train_mod)\n",
    "clean_digit = knn_clf.predict([x_test_mod[120]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(x_test_mod[120].reshape(28,28), cmap='binary');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(clean_digit.reshape(28,28),cmap='binary');"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "63e79917a05e390872358bfb73c58bc903ada01d2d04077091749088207d82cf"
  },
  "kernelspec": {
   "display_name": "Python 3.10.0 64-bit",
   "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": "3.10.0"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}