malaria_classification_knn.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 20,
      "metadata": {
        "pycharm": {
          "is_executing": false
        }
      },
      "outputs": [],
      "source": "import cv2\nimport os\nfrom sklearn.model_selection import train_test_split\nimport numpy as np\nimport timeit\nfrom PIL import Image\nfrom sklearn.preprocessing import StandardScaler\nfrom sklearn.neighbors import KNeighborsClassifier\nfrom sklearn.metrics import confusion_matrix, accuracy_score, recall_score\nfrom sklearn.metrics import classification_report\nfrom sklearn.model_selection import GridSearchCV\nfrom sklearn.feature_selection import VarianceThreshold\nfrom sklearn.pipeline import Pipeline\nimport matplotlib.pyplot as plt\n"
    },
    {
      "cell_type": "markdown",
      "source": "### Data Preparation",
      "metadata": {
        "pycharm": {
          "metadata": false,
          "name": "#%% md\n"
        }
      }
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "pycharm": {}
      },
      "source": "#### create generator of infected and uninfected images"
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "pycharm": {
          "is_executing": false,
          "metadata": false,
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": "infected \u003d (image for image in os.listdir(\u0027Parasitized\u0027))\nuninfected \u003d (image for image in os.listdir(\u0027Uninfected\u0027))\ndata \u003d []\nlabels \u003d []\n"
    },
    {
      "cell_type": "markdown",
      "source": "#### function to load data from previous run",
      "metadata": {
        "pycharm": {
          "metadata": false,
          "name": "#%% md\n"
        }
      }
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "outputs": [],
      "source": "def load_data():\n    \"\"\" load cell and label data from disk \"\"\"\n    cells \u003d np.load(\u0027data/Cells.npy\u0027)\n    labels \u003d np.load(\u0027data/Labels.npy\u0027)\n    return cells, labels\n",
      "metadata": {
        "pycharm": {
          "metadata": false,
          "name": "#%%\n",
          "is_executing": false
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": "#### run this block if loading cell and label data from previous run",
      "metadata": {
        "pycharm": {
          "metadata": false
        }
      }
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "outputs": [],
      "source": "cells, labels \u003d load_data()\n",
      "metadata": {
        "pycharm": {
          "metadata": false,
          "name": "#%%\n",
          "is_executing": false
        }
      }
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "pycharm": {}
      },
      "source": "#### convert images to grayscale, and resize to 50x50 "
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "pycharm": {
          "is_executing": false,
          "metadata": false,
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": "def transform_images():\n    \"\"\" convert color images to grayscale, resize\"\"\"\n    for img in infected:\n        # image \u003d cv2.imread(\u0027./Parasitized/\u0027 + img)\n        image \u003d cv2.imread(\u0027./Parasitized/\u0027 + img, cv2.IMREAD_GRAYSCALE)\n        image_array \u003d Image.fromarray(image)  # create PIL image\n        resize_img \u003d image_array.resize((50, 50))  # returns resized copy of image\n        \"\"\"\n        blurring the image reduces noise \n        Blurs/smooths an image using the normalized box filter\n        \"\"\"\n        # blur_image \u003d cv2.blur(np.array(resize_img), (50,50))\n        data.append(np.array(resize_img))\n        labels.append(1)\n\n    for img in uninfected:\n        # image \u003d cv2.imread(\u0027./Uninfected/\u0027 + img)\n        image \u003d cv2.imread(\u0027./Uninfected/\u0027 + img, cv2.IMREAD_GRAYSCALE)\n        image_array \u003d Image.fromarray(image)  # create PIL image\n        resize_img \u003d image_array.resize((50, 50))  # returns resized copy of image\n        # blur_image \u003d cv2.blur(np.array(resize_img), (50,50))\n        data.append(np.array(resize_img))\n        labels.append(0)\n        \ntransform_images()\n"
    },
    {
      "cell_type": "markdown",
      "source": "#### convert to numpy ndarray",
      "metadata": {
        "pycharm": {
          "metadata": false
        }
      }
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "outputs": [],
      "source": "cells \u003d np.array(data).astype(\u0027float64\u0027)\nlabels \u003d np.array(labels).astype(\u0027float64\u0027)\n",
      "metadata": {
        "pycharm": {
          "metadata": false,
          "name": "#%%\n",
          "is_executing": false
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": "#### inspect images",
      "metadata": {
        "pycharm": {
          "metadata": false
        }
      }
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "outputs": [
        {
          "data": {
            "text/plain": "\u003cFigure size 360x360 with 26 Axes\u003e",
            "image/png": "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\u003d\u003d\n"
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": "def inspect_images():\n    np.random.seed(155)\n    random_25 \u003d np.random.randint(0,cells.shape[0], size\u003d25)\n    fig\u003dplt.figure(figsize\u003d(5,5))\n    plt.title(\"Infected/Uninfected\",fontsize\u003d20)\n    plt.axis(\u0027off\u0027)\n    for ind,rand in enumerate(random_25):\n        ax_n\u003dfig.add_subplot(5,5,ind+1)\n        ax_n.imshow(cells[rand],cmap\u003dplt.cm.gray)\n        ax_n.get_xaxis().set_visible(False)\n        ax_n.get_yaxis().set_visible(False)\n    plt.show()\ninspect_images()\n",
      "metadata": {
        "pycharm": {
          "metadata": false,
          "name": "#%% \n",
          "is_executing": false
        }
      }
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "pycharm": {}
      },
      "source": "#### save ndarray to disk"
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "pycharm": {
          "is_executing": false,
          "metadata": false,
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": "np.save(\u0027Cells\u0027, cells, allow_pickle\u003dFalse) \nnp.save(\u0027Labels\u0027, labels, allow_pickle\u003dFalse) \n"
    },
    {
      "cell_type": "code",
      "execution_count": 44,
      "metadata": {
        "pycharm": {
          "is_executing": false,
          "metadata": false,
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "text": [
            "Cells : (27558, 50, 50) | labels : (27558,)\n"
          ],
          "output_type": "stream"
        }
      ],
      "source": "print(\u0027Cells : {} | labels : {}\u0027.format(cells.shape , labels.shape))\n"
    },
    {
      "cell_type": "markdown",
      "source": "#### flatten the 50x50x3 or 50x50 dimensions into a feature vector of 7500 or 2500 values \n+ demensions for color images are 50x50x3 and 50x50 for grayscale",
      "metadata": {
        "pycharm": {
          "metadata": false
        }
      }
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "outputs": [],
      "source": "# cells_flattened \u003d cells[:].reshape(-1,7500)\ncells_flattened \u003d cells[:].reshape(27558,2500)",
      "metadata": {
        "pycharm": {
          "metadata": false,
          "name": "#%%\n",
          "is_executing": false
        }
      }
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "outputs": [
        {
          "data": {
            "text/plain": "(27558, 2500)"
          },
          "metadata": {},
          "output_type": "execute_result",
          "execution_count": 17
        }
      ],
      "source": "cells_flattened.shape\n",
      "metadata": {
        "pycharm": {
          "metadata": false,
          "name": "#%%\n",
          "is_executing": false
        }
      }
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "pycharm": {}
      },
      "source": "#### create train/test dataset"
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "pycharm": {
          "is_executing": false,
          "metadata": false,
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": "x_train, x_validation, y_train, y_validation \u003d train_test_split(cells_flattened, labels, train_size\u003d0.9, test_size\u003d0.1, random_state\u003d155)\n"
    },
    {
      "cell_type": "markdown",
      "source": "### Use grid search to find best hyperparameters for KNN",
      "metadata": {
        "pycharm": {
          "metadata": false,
          "name": "#%% md\n"
        }
      }
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "pycharm": {}
      },
      "source": "#### GridSearchCV wrapper implements cross validation and scoring metric\n+ https://scikit-learn.org/stable/auto_examples/model_selection/plot_grid_search_digits.html"
    },
    {
      "cell_type": "code",
      "execution_count": 21,
      "metadata": {
        "pycharm": {
          "is_executing": false,
          "metadata": false,
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": "def grid_search(X, y):\n    standardized_data \u003d StandardScaler()\n    var_filter \u003d VarianceThreshold()\n    knn \u003d KNeighborsClassifier()\n    pipe \u003d Pipeline([(\u0027standardized_data\u0027, standardized_data),\n                     (\u0027var_filter\u0027, var_filter),\n                     (\u0027knn\u0027, knn)])\n    \n    parameters \u003d [{\u0027knn__n_neighbors\u0027:[1,2,3,4,5], \u0027knn__algorithm\u0027:[\u0027brute\u0027], \u0027knn__metric\u0027:[\u0027euclidean\u0027]}]\n    grid_search \u003d GridSearchCV(pipe, parameters, cv\u003d5, n_jobs\u003d4, scoring\u003d\u0027accuracy\u0027)\n    \n    start \u003d timeit.default_timer()\n    grid_search.fit(X, y)\n    print(f\u0027Runtime for KNN:{timeit.default_timer() - start}\u0027)\n    return grid_search\n"
    },
    {
      "cell_type": "markdown",
      "source": "#### get best model returned from grid search",
      "metadata": {
        "pycharm": {
          "metadata": false,
          "name": "#%% md\n"
        }
      }
    },
    {
      "cell_type": "code",
      "execution_count": 22,
      "outputs": [
        {
          "name": "stdout",
          "text": [
            "Runtime for KNN:487.17083713500006\n"
          ],
          "output_type": "stream"
        }
      ],
      "source": "knn_model \u003d grid_search(x_train, y_train) \n",
      "metadata": {
        "pycharm": {
          "metadata": false,
          "name": "#%%\n",
          "is_executing": false
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": "#### get best model and mean score from grid search ",
      "metadata": {
        "pycharm": {
          "metadata": false,
          "name": "#%% md\n"
        }
      }
    },
    {
      "cell_type": "code",
      "execution_count": 23,
      "outputs": [
        {
          "name": "stdout",
          "text": [
            "{\u0027knn__algorithm\u0027: \u0027brute\u0027, \u0027knn__metric\u0027: \u0027euclidean\u0027, \u0027knn__n_neighbors\u0027: 1}\n"
          ],
          "output_type": "stream"
        },
        {
          "data": {
            "text/plain": "array([0.58563825, 0.54185146, 0.57330054, 0.54769777, 0.56870414])"
          },
          "metadata": {},
          "output_type": "execute_result",
          "execution_count": 23
        }
      ],
      "source": "print(knn_model.best_params_)\nknn_model.cv_results_[\u0027mean_test_score\u0027]\n    ",
      "metadata": {
        "pycharm": {
          "metadata": false,
          "name": "#%% \n",
          "is_executing": false
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": "### Performance metrics",
      "metadata": {
        "pycharm": {
          "metadata": false,
          "name": "#%% md\n"
        }
      }
    },
    {
      "cell_type": "code",
      "execution_count": 27,
      "metadata": {
        "pycharm": {
          "is_executing": false,
          "metadata": false,
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": "results \u003d knn_model.predict(x_validation)"
    },
    {
      "cell_type": "code",
      "execution_count": 28,
      "metadata": {
        "pycharm": {
          "is_executing": false,
          "metadata": false,
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": "array([[1209,  188],\n       [ 935,  424]])"
          },
          "metadata": {},
          "output_type": "execute_result",
          "execution_count": 28
        }
      ],
      "source": "confusion_matrix(y_validation, results)"
    },
    {
      "cell_type": "code",
      "execution_count": 29,
      "metadata": {
        "pycharm": {
          "is_executing": false,
          "metadata": false,
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "text": [
            "Accuracy: 0.5925253991291727\nRecall: 0.31199411331861665\n              precision    recall  f1-score   support\n\n           0       0.56      0.87      0.68      1397\n           1       0.69      0.31      0.43      1359\n\n   micro avg       0.59      0.59      0.59      2756\n   macro avg       0.63      0.59      0.56      2756\nweighted avg       0.63      0.59      0.56      2756\n\n"
          ],
          "output_type": "stream"
        }
      ],
      "source": "print(f\"Accuracy: {accuracy_score(y_validation, results)}\")\nprint(f\"Recall: {recall_score(y_validation, results)}\")\nprint(classification_report(y_validation, results))"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "outputs": [],
      "source": "\n",
      "metadata": {
        "pycharm": {
          "metadata": false,
          "name": "#%%\n"
        }
      }
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.7.2"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 1
}