K_neariest_neighbor.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "gjX47FjcFRG1"
   },
   "source": [
    "# Part 1. Implementing [KNN](https://en.wikipedia.org/wiki/K-nearest_neighbors_algorithm) with numpy."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "RB7EfXpgvhrr"
   },
   "outputs": [],
   "source": [
    "# Hint: np.newaxis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'3.9.12 (main, Apr  4 2022, 05:22:27) [MSC v.1916 64 bit (AMD64)]'"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import os\n",
    "os.sys.version"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ykY2QFGGZDdV"
   },
   "source": [
    "## K-neariest neighbor classifier\n",
    "\n",
    "You are implementing one of the simplest machine learning algorithms, a K-Nearest Neighbor Classifier.\n",
    "We will apply it to tasks:\n",
    "- binary classification (that is, only two classes)\n",
    "- multi-class classification (that is, multiple classes)\n",
    "\n",
    "Since the method needs a hyperparameter - the number of neighbors, we will select it based on cross-validation.\n",
    "\n",
    "Before completing the assignment:\n",
    "- Put the files 'knn.py', 'metrics.py', 'dataset.py', `download_data.sh` into the current directory (/ content /)\n",
    "- run the file `download_data.sh` to download the data that we will use for training into the data directory (data will appear during the launch of the bash script)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "id": "osS7LAzmZdkY"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "mkdir: cannot create directory â€˜dataâ€™: File exists\n",
      "--2022-10-27 12:21:53--  http://ufldl.stanford.edu/housenumbers/train_32x32.mat\n",
      "Resolving ufldl.stanford.edu (ufldl.stanford.edu)... 171.64.68.10\n",
      "Connecting to ufldl.stanford.edu (ufldl.stanford.edu)|171.64.68.10|:80... connected.\n",
      "HTTP request sent, awaiting response... 416 Requested Range Not Satisfiable\n",
      "\n",
      "    The file is already fully retrieved; nothing to do.\n",
      "\n",
      "--2022-10-27 12:21:54--  http://ufldl.stanford.edu/housenumbers/test_32x32.mat\n",
      "Reusing existing connection to ufldl.stanford.edu:80.\n",
      "HTTP request sent, awaiting response... 416 Requested Range Not Satisfiable\n",
      "\n",
      "    The file is already fully retrieved; nothing to do.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "! bash download_data.sh"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "id": "DPe5pDJ0ZDdZ"
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "%matplotlib inline\n",
    "\n",
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "id": "Wymy4bozZDda"
   },
   "outputs": [],
   "source": [
    "from dataset import load_svhn\n",
    "from knn import KNN\n",
    "from metrics import binary_classification_metrics, multiclass_accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "TkBIQCI1ZDdb"
   },
   "source": [
    "## Load and visualize data\n",
    "\n",
    "The task has already given the `load_svhn` function, which loads data from disk. It returns data for training and testing as numpy arrays.\n",
    "\n",
    "We'll use numbers from the Street View House Numbers dataset (SVHN, http://ufldl.stanford.edu/housenumbers/)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 194,
   "metadata": {
    "id": "_jPml2nCZDdc"
   },
   "outputs": [],
   "source": [
    "train_X, train_y, test_X, test_y = load_svhn(\"data\", max_train=1000, max_test=100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 239
    },
    "id": "wJnejYxVZDdc",
    "outputId": "31348fa7-bfe0-4f2e-bef6-7d4bed632108"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 50 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "samples_per_class = 5  # Number of samples per class to visualize\n",
    "plot_index = 1\n",
    "for example_index in range(samples_per_class):\n",
    "    for class_index in range(10):\n",
    "        plt.subplot(5, 10, plot_index)\n",
    "        image = train_X[train_y == class_index][example_index]\n",
    "        plt.imshow(image.astype(np.uint8))\n",
    "        plt.axis('off')\n",
    "        plot_index += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "SAygqoKfZDdd"
   },
   "source": [
    "## First we implement KNN for binary classification\n",
    "\n",
    "As a binary classification task, we will train a model to distinguish the number 0 from the number 9."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "id": "Qn5LXNMIZDde"
   },
   "outputs": [],
   "source": [
    "# First, let's prepare the labels and the source data\n",
    "\n",
    "# Only select 0s and 9s\n",
    "binary_train_mask = (train_y == 0) | (train_y == 9)\n",
    "binary_train_X = train_X[binary_train_mask]\n",
    "binary_train_y = train_y[binary_train_mask] == 0\n",
    "\n",
    "binary_test_mask = (test_y == 0) | (test_y == 9)\n",
    "binary_test_X = test_X[binary_test_mask]\n",
    "binary_test_y = test_y[binary_test_mask] == 0\n",
    "\n",
    "# Reshape to 1-dimensional array [num_samples, 32*32*3]\n",
    "binary_train_X = binary_train_X.reshape(binary_train_X.shape[0], -1)\n",
    "binary_test_X = binary_test_X.reshape(binary_test_X.shape[0], -1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "id": "f6NkPMHwZDdf"
   },
   "outputs": [],
   "source": [
    "# Create the classifier and call fit to train the model\n",
    "# KNN just remembers all the data\n",
    "knn_classifier = KNN(k=1, metric='manhattan')\n",
    "knn_classifier.fit(binary_train_X, binary_train_y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "JIOgIArLZDdg"
   },
   "source": [
    "### It's time to write some code!\n",
    "\n",
    "Implement the functions `compute_distances_two_loops`,` compute_distances_one_loop` and `compute_distances_no_loops`\n",
    "in the file `knn.py`.\n",
    "\n",
    "These functions build an array of distances between all vectors in the test set and in the training set.\n",
    "As a result, they must build an array of size `(num_test, num_train)`, where the coordinate `[i][j]` corresponds to the distance between the i-th vector in test (`test[i]`) and the j-th vector in train ( `train[j]`).\n",
    "\n",
    "**Please note** We will use the measures L1 and L2 as a distance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "id": "PCynlyn_ZDdh"
   },
   "outputs": [],
   "source": [
    "# TODO: implement compute_distances_two_loops in knn.py\n",
    "dists = knn_classifier.compute_distances_two_loops(binary_test_X)\n",
    "assert np.isclose(dists[0, 10], np.sum(np.abs(binary_test_X[0] - binary_train_X[10])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "id": "0EIUa71yZDdi"
   },
   "outputs": [],
   "source": [
    "# TODO: implement compute_distances_one_loop in knn.py\n",
    "dists = knn_classifier.compute_distances_one_loop(binary_test_X)\n",
    "assert np.isclose(dists[0, 10], np.sum(np.abs(binary_test_X[0] - binary_train_X[10])))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "id": "QJk3FT-KZDdi"
   },
   "outputs": [],
   "source": [
    "# TODO: implement compute_distances_no_loops in knn.py\n",
    "dists = knn_classifier.compute_distances_no_loops(binary_test_X)\n",
    "assert np.isclose(dists[0, 10], np.sum(np.abs(binary_test_X[0] - binary_train_X[10])))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "knn_classifier = KNN(k=1, metric='euclidean')\n",
    "knn_classifier.fit(binary_train_X, binary_train_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TODO: implement compute_distances_two_loops in knn.py euclidean metrics\n",
    "dists = knn_classifier.compute_distances_two_loops(binary_test_X)\n",
    "assert np.isclose(dists[0, 10], np.sqrt(np.sum(np.square(binary_test_X[0] - binary_train_X[10]))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TODO: implement compute_distances_one_loop in knn.py euclidean metrics\n",
    "dists = knn_classifier.compute_distances_one_loop(binary_test_X)\n",
    "assert np.isclose(dists[0, 10], np.sqrt(np.sum(np.square(binary_test_X[0] - binary_train_X[10]))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TODO: implement compute_distances_no_loops in knn.py euclidean metrics\n",
    "dists = knn_classifier.compute_distances_no_loops(binary_test_X)\n",
    "assert np.isclose(dists[0, 10], np.sqrt(np.sum(np.square(binary_test_X[0] - binary_train_X[10]))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "knn_classifier = KNN(k=1, metric='manhattan')\n",
    "knn_classifier.fit(binary_train_X, binary_train_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "id": "smJtS-j4ZDdj"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "37.8 ms ± 2.51 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "40.7 ms ± 1.38 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "39.2 ms ± 1.27 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n"
     ]
    }
   ],
   "source": [
    "# Lets look at the performance difference\n",
    "%timeit knn_classifier.compute_distances_two_loops(binary_test_X)\n",
    "%timeit knn_classifier.compute_distances_one_loop(binary_test_X)\n",
    "%timeit knn_classifier.compute_distances_no_loops(binary_test_X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "id": "UksPS6D3ZDdk"
   },
   "outputs": [],
   "source": [
    "# TODO: implement predict_labels_binary in knn.py\n",
    "prediction = knn_classifier.predict(binary_test_X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "id": "a-D_IOHnZDdl"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "KNN with k = 1\n",
      "Accuracy: 0.56, Precision: 0.73, Recall: 0.67, F1: 0.70\n"
     ]
    }
   ],
   "source": [
    "# TODO: implement binary_classification_metrics in metrics.py\n",
    "precision, recall, f1, accuracy = binary_classification_metrics(prediction, binary_test_y)\n",
    "print(\"KNN with k = %s\" % knn_classifier.k)\n",
    "print(\"Accuracy: %4.2f, Precision: %4.2f, Recall: %4.2f, F1: %4.2f\" % (accuracy, precision, recall, f1)) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {
    "id": "KZ0foAIqZDdl"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "KNN with k = 3\n",
      "Accuracy: 0.56, Precision: 0.73, Recall: 0.67, F1: 0.70\n"
     ]
    }
   ],
   "source": [
    "# Let's put everything together and run KNN with k=3 and see how we do\n",
    "knn_classifier_3 = KNN(k=3,metric='manhattan')\n",
    "knn_classifier_3.fit(binary_train_X, binary_train_y)\n",
    "prediction = knn_classifier_3.predict(binary_test_X)\n",
    "\n",
    "precision, recall, f1, accuracy = binary_classification_metrics(prediction, binary_test_y)\n",
    "print(\"KNN with k = %s\" % knn_classifier_3.k)\n",
    "print(\"Accuracy: %4.2f, Precision: %4.2f, Recall: %4.2f, F1: %4.2f\" % (accuracy, precision, recall, f1)) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ysuUKK-HZDdl"
   },
   "source": [
    "## Cross-validation\n",
    "\n",
    "Let's try to find the best value for the parameter k for the KNN algorithm!\n",
    "\n",
    "For this we will use k-fold cross-validation (https://en.wikipedia.org/wiki/Cross-validation_(statistics)#k-fold_cross-validation). We'll split the training data into 5 folds and in turn use each of them as validation data and the rest as training data.\n",
    "\n",
    "As a final estimate of the effectiveness of k, we will average the F1 score across all folds.\n",
    "After that, we'll just pick the k value with the best metric value."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 227,
   "metadata": {
    "id": "JsrXQVFmZDdm"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "k = 1, f1 = 0.820000\n",
      "k = 2, f1 = 0.570000\n",
      "k = 3, f1 = 0.680000\n",
      "k = 5, f1 = 0.660000\n",
      "k = 8, f1 = 0.740000\n",
      "k = 10, f1 = 0.620000\n",
      "k = 15, f1 = 0.620000\n",
      "k = 20, f1 = 0.590000\n",
      "k = 25, f1 = 0.600000\n",
      "k = 50, f1 = 0.650000\n"
     ]
    }
   ],
   "source": [
    "# Find the best k using cross-validation based on F1 score\n",
    "import random\n",
    "num_folds = 5\n",
    "\n",
    "\n",
    "# TODO: split the training data in 5 folds and store them in train_folds_X/train_folds_y\n",
    "\n",
    "k_choices = [1, 2, 3, 5, 8, 10, 15, 20, 25, 50]\n",
    "k_to_f1 = {}  # dict mapping k values to mean F1 scores (int -> float)\n",
    "\n",
    "fold_size=int(binary_train_X.shape[0]/num_folds)\n",
    "\n",
    "for k in k_choices:\n",
    "    # TODO: perform cross-validation\n",
    "    # Go through every fold and use it for testing and all other folds for training\n",
    "    # Perform training and produce F1 score metric on the validation dataset\n",
    "    # Average F1 from all the folds and write it into k_to_f1\n",
    "    train_folds_X = []\n",
    "    train_folds_y = []\n",
    "    for i in range(num_folds):\n",
    "        fold_X=[]\n",
    "        fold_y=[]\n",
    "        while len(fold_X) < fold_size:\n",
    "            index=random.randrange(binary_train_X.shape[0])\n",
    "            fold_X.append(binary_train_X[index])\n",
    "            fold_y.append(binary_train_y[index])\n",
    "        train_folds_X.append(fold_X)\n",
    "        train_folds_y.append(fold_y)\n",
    "    \n",
    "    F1=0\n",
    "    average_f1=0\n",
    "    for fold in range(num_folds):\n",
    "        valid_X=train_folds_X[fold]\n",
    "        \n",
    "        len_val=len(train_folds_X[fold])\n",
    "        valid_X=np.concatenate(valid_X,axis=0)\n",
    "        valid_X=valid_X.reshape(len_val,int(len(valid_X)/len_val))\n",
    "        \n",
    "        valid_y=train_folds_y[fold]\n",
    "        \n",
    "        cross_X=np.delete(train_folds_X,fold,axis=0)\n",
    "        cross_X=cross_X.reshape(cross_X.shape[0]*cross_X.shape[1],cross_X.shape[2])\n",
    "        \n",
    "        cross_y=np.delete(train_folds_y,fold,axis=0).reshape(-1)\n",
    "        \n",
    "        knn_classifier = KNN(k=k,metric='manhattan')\n",
    "        knn_classifier.fit(cross_X, cross_y)\n",
    "        prediction = knn_classifier.predict(valid_X)\n",
    "    \n",
    "        precision, recall, f1, accuracy = binary_classification_metrics(prediction, valid_y)\n",
    "        \n",
    "        F1+=f1\n",
    "    average_f1=round(F1/num_folds,2)\n",
    "    \n",
    "    k_to_f1[k]=average_f1\n",
    "\n",
    "for k in sorted(k_to_f1):\n",
    "    print('k = %d, f1 = %f' % (k, k_to_f1[k]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "g3elyRwpZDdn"
   },
   "source": [
    "### Check how well the best k value works on test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "metadata": {
    "id": "kTaaQ1siZDdn"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best KNN with k = 1\n",
      "Accuracy: 0.56, Precision: 0.73, Recall: 0.67, F1: 0.70\n"
     ]
    }
   ],
   "source": [
    "# TODO Set the best k to the best value found by cross-validation\n",
    "best_k = 1\n",
    "\n",
    "best_knn_classifier = KNN(k=best_k,metric='manhattan')\n",
    "best_knn_classifier.fit(binary_train_X, binary_train_y)\n",
    "prediction = best_knn_classifier.predict(binary_test_X)\n",
    "\n",
    "precision, recall, f1, accuracy = binary_classification_metrics(prediction, binary_test_y)\n",
    "print(\"Best KNN with k = %s\" % best_k)\n",
    "print(\"Accuracy: %4.2f, Precision: %4.2f, Recall: %4.2f, F1: %4.2f\" % (accuracy, precision, recall, f1)) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "yKqM0zukZDdn"
   },
   "source": [
    "## multi-class classification\n",
    "\n",
    "We pass to the next stage - classification for each digit."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "metadata": {
    "id": "cTayszopZDdo"
   },
   "outputs": [],
   "source": [
    "# Now let's use all 10 classes\n",
    "train_X = train_X.reshape(train_X.shape[0], -1)\n",
    "test_X = test_X.reshape(test_X.shape[0], -1)\n",
    "\n",
    "knn_classifier = KNN(k=1,metric='manhattan')\n",
    "knn_classifier.fit(train_X, train_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 202,
   "metadata": {
    "id": "S-FZz10hZDdo"
   },
   "outputs": [],
   "source": [
    "# TODO: Implement predict_labels_multiclass\n",
    "predict = knn_classifier.predict(test_X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 206,
   "metadata": {
    "id": "ANCNzvs8ZDdo"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.21\n"
     ]
    }
   ],
   "source": [
    "# TODO: Implement multiclass_accuracy\n",
    "accuracy = multiclass_accuracy(predict, test_y)\n",
    "print(\"Accuracy: %4.2f\" % accuracy)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "vx-ME76VZDdp"
   },
   "source": [
    "Cross-validation again. Now our main metric is accuracy, and we will also average it over all folds."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 235,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "k = 1, accuracy = 0.670000\n",
      "k = 2, accuracy = 0.530000\n",
      "k = 3, accuracy = 0.450000\n",
      "k = 5, accuracy = 0.390000\n",
      "k = 8, accuracy = 0.320000\n",
      "k = 10, accuracy = 0.320000\n",
      "k = 15, accuracy = 0.330000\n",
      "k = 20, accuracy = 0.290000\n",
      "k = 25, accuracy = 0.270000\n",
      "k = 50, accuracy = 0.250000\n"
     ]
    }
   ],
   "source": [
    "# Find the best k using cross-validation based on F1 score\n",
    "import random\n",
    "num_folds = 5\n",
    "\n",
    "\n",
    "# TODO: split the training data in 5 folds and store them in train_folds_X/train_folds_y\n",
    "\n",
    "k_choices = [1, 2, 3, 5, 8, 10, 15, 20, 25, 50]\n",
    "k_to_accuracy = {}  \n",
    "\n",
    "fold_size=int(train_X.shape[0]/num_folds)\n",
    "\n",
    "for k in k_choices:\n",
    "    # TODO: perform cross-validation\n",
    "    # Go through every fold and use it for testing and all other folds for training\n",
    "    # Perform training and produce F1 score metric on the validation dataset\n",
    "    # Average F1 from all the folds and write it into k_to_f1\n",
    "    train_folds_X = []\n",
    "    train_folds_y = []\n",
    "    for i in range(num_folds):\n",
    "        fold_X=[]\n",
    "        fold_y=[]\n",
    "        while len(fold_X) < fold_size:\n",
    "            index=random.randrange(train_X.shape[0])\n",
    "            fold_X.append(train_X[index])\n",
    "            fold_y.append(train_y[index])\n",
    "        train_folds_X.append(fold_X)\n",
    "        train_folds_y.append(fold_y)\n",
    "    \n",
    "    sum_acc=0\n",
    "    average_acc=0\n",
    "    for fold in range(num_folds):\n",
    "        valid_X=train_folds_X[fold]\n",
    "        \n",
    "        len_val=len(train_folds_X[fold])\n",
    "        valid_X=np.concatenate(valid_X,axis=0)\n",
    "        valid_X=valid_X.reshape(len_val,int(len(valid_X)/len_val))\n",
    "        \n",
    "        valid_y=train_folds_y[fold]\n",
    "        \n",
    "        cross_X=np.delete(train_folds_X,fold,axis=0)\n",
    "        cross_X=cross_X.reshape(cross_X.shape[0]*cross_X.shape[1],cross_X.shape[2])\n",
    "        \n",
    "        cross_y=np.delete(train_folds_y,fold,axis=0).reshape(-1)\n",
    "        \n",
    "        knn_classifier = KNN(k=k,metric='manhattan')\n",
    "        knn_classifier.fit(cross_X, cross_y)\n",
    "        prediction = knn_classifier.predict(valid_X)\n",
    "    \n",
    "        accuracy = multiclass_accuracy(prediction, valid_y)\n",
    "        \n",
    "        sum_acc+=accuracy\n",
    "    average_acc=round(sum_acc/num_folds,2)\n",
    "    \n",
    "    k_to_accuracy[k]=average_acc\n",
    "\n",
    "for k in sorted(k_to_accuracy):\n",
    "    print('k = %d, accuracy = %f' % (k, k_to_accuracy[k]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "wQXVwaTsZDdp"
   },
   "source": [
    "### Final test - classification into 10 classes on the test sample (test data)\n",
    "\n",
    "If implemented correctly, you should see an accuracy of at least **0.2**."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 236,
   "metadata": {
    "id": "XSGCjoZ6ZDdq"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.21\n"
     ]
    }
   ],
   "source": [
    "# TODO Set the best k as a best from computed\n",
    "best_k = 1\n",
    "\n",
    "best_knn_classifier = KNN(k=best_k, metric='manhattan')\n",
    "best_knn_classifier.fit(train_X, train_y)\n",
    "prediction = best_knn_classifier.predict(test_X)\n",
    "\n",
    "# Accuracy should be around 20%!\n",
    "accuracy = multiclass_accuracy(prediction, test_y)\n",
    "print(\"Accuracy: %4.2f\" % accuracy)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "15756ZYT4XnQ"
   },
   "source": [
    "### Now it remains to push all 5 files into the repository, and attach the link to your repository to the classroom ... Or, before that, you can try to quickly implement this with sklearn!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "rLzxgoLQFkYk"
   },
   "source": [
    "# Part 2 (Bonus). Sklearn "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "i8KmqDMkGVA3"
   },
   "source": [
    "Compare your models for both metrics with similar ones in sklearn."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.neighbors import KNeighborsClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 240,
   "metadata": {},
   "outputs": [],
   "source": [
    "knn=KNeighborsClassifier(n_neighbors=1,metric='manhattan')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 241,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "KNeighborsClassifier(metric='manhattan', n_neighbors=1)"
      ]
     },
     "execution_count": 241,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "knn.fit(train_X,train_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 242,
   "metadata": {},
   "outputs": [],
   "source": [
    "prediction=knn.predict(test_X)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 243,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import accuracy_score\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 244,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.21"
      ]
     },
     "execution_count": 244,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(test_y,prediction)\n"
   ]
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "K-neariest neighbor",
   "provenance": [],
   "toc_visible": true
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}