Malaria_Detection_CNN.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 224
    },
    "colab_type": "code",
    "id": "qAgD6vb1-vdf",
    "outputId": "23d0eece-fdf2-410d-f7ff-d9b719b2e594"
   },
   "outputs": [],
   "source": [
    "# !wget --header=\"Host: storage.googleapis.com\" --header=\"User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_14_3) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/73.0.3683.86 Safari/537.36\" --header=\"Accept: text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3\" --header=\"Accept-Language: en-US,en;q=0.9\" --header=\"Referer: https://www.kaggle.com/\" \"https://storage.googleapis.com/kaggle-datasets/87153/200743/cell-images-for-detecting-malaria.zip?GoogleAccessId=web-data@kaggle-161607.iam.gserviceaccount.com&Expires=1554030372&Signature=TebHa3x60W5Fh8Eh09X7a6q6eHQgfJI15kSz1jzQ%2FbPKOJ8K%2BpLtM%2BfEsXbu%2BpW0ByIFgn49BV844ketChZozyziW8tByVJ2f3MXDxP5HAQH1b8z226UCeLsRQZR7eWKgH25A24D1W5HopBwhRH0sC7QiXksNG6fMeQrI0gQp2MsPmJXB1m4yQlRx%2BCc2o2VajVK2wuMD%2FjS%2BSrKcDH8NxUHfRMJJDIh4loV3mqe6Cn2BSRfUwl4G4EOpoqbv%2BxMNXb0FgOPv7yVqoQkYHfYSHa8Q34RY7Nndlkow53wpofVkjB9BxgyZZpMekplst63vHH84fzaZcbWUANn5s92Rg%3D%3D\" -O \"cell-images-for-detecting-malaria.zip\" -c\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "4zeoBAGW-_ix"
   },
   "outputs": [],
   "source": [
    "# !unzip cell-images-for-detecting-malaria.zip"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "BTuc8PQX_G_i",
    "outputId": "13c063a3-8759-4ea7-c8e9-e8d19e1bed51"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parasitized  Uninfected\r\n"
     ]
    }
   ],
   "source": [
    "!ls cell_images/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "YhLUrbsy_L7c"
   },
   "outputs": [],
   "source": [
    "%reload_ext autoreload\n",
    "%autoreload 2\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "0CCtvCAv8gts"
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from PIL import Image\n",
    "import os\n",
    "\n",
    "import torch\n",
    "from torch import nn, optim\n",
    "from torchvision import transforms, datasets, models\n",
    "from torch.utils.data.sampler import SubsetRandomSampler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "from my_ml_lib import MetricTools, PlotTools"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "rwWVAKjnTNNv"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "device(type='cuda', index=0)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
    "device"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "K83Yo1D2_3Yn"
   },
   "outputs": [],
   "source": [
    "data_transforms = transforms.Compose([\n",
    "    transforms.RandomResizedCrop(64),\n",
    "    transforms.ToTensor(),\n",
    "    transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "CiiEfPGB_d7K"
   },
   "outputs": [],
   "source": [
    "data_dir = 'cell_images/'\n",
    "data = datasets.ImageFolder(data_dir, transform=data_transforms)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "CU5KSsoMK2t2"
   },
   "outputs": [],
   "source": [
    "valid_pct = 0.2\n",
    "test_pct = 0.1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "pyE5Gbg3KRg2",
    "outputId": "e860ca95-6930-4f0b-dcbb-f5c0eb1df969"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5511 2756 19291\n"
     ]
    }
   ],
   "source": [
    "data_len = len(data)\n",
    "data_indices = list(range(data_len))\n",
    "np.random.shuffle(data_indices)\n",
    "\n",
    "valid_split = int(np.floor((valid_pct) * data_len))\n",
    "test_split = int(np.floor((valid_pct + test_pct) * data_len))\n",
    "\n",
    "valid_idx, test_idx, train_idx = data_indices[:valid_split], data_indices[valid_split:test_split], data_indices[test_split:]\n",
    "\n",
    "train_sampler = SubsetRandomSampler(train_idx)\n",
    "valid_sampler = SubsetRandomSampler(valid_idx)\n",
    "test_sampler = SubsetRandomSampler(test_idx)\n",
    "\n",
    "dataloaders = {\n",
    "    'train': torch.utils.data.DataLoader(data, batch_size=64, sampler=train_sampler),\n",
    "    'valid': torch.utils.data.DataLoader(data, batch_size=32, sampler=valid_sampler),\n",
    "    'test': torch.utils.data.DataLoader(data, batch_size=20, sampler=test_sampler)\n",
    "}\n",
    "\n",
    "print(len(valid_idx), len(test_idx), len(train_idx))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "nLIBfFgZMYcC"
   },
   "outputs": [],
   "source": [
    "invTrans = transforms.Compose([ transforms.Normalize([ 0., 0., 0. ],[ 1/0.229, 1/0.224, 1/0.225 ]),\n",
    "                                transforms.Normalize([ -0.485, -0.456, -0.406 ],[ 1., 1., 1. ]),\n",
    "                               ])\n",
    "\n",
    "def imshow(img):\n",
    "    img = invTrans(img)\n",
    "    img = img.numpy()\n",
    "    plt.imshow(np.transpose(img, (1, 2, 0)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 372
    },
    "colab_type": "code",
    "id": "5n_S8EUaQD3T",
    "outputId": "b63e900f-d396-4ec2-f699-6a8fa98422db"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x432 with 9 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "classes=['infected','uninfected']\n",
    "\n",
    "dataiter = iter(dataloaders['train'])\n",
    "images, labels = dataiter.next()\n",
    "\n",
    "fig = plt.figure(figsize=(6, 6))\n",
    "# display 20 images\n",
    "for idx in np.arange(9):\n",
    "    ax = fig.add_subplot(3, 3, idx+1, xticks=[], yticks=[])\n",
    "    imshow(images[idx])\n",
    "    ax.set_title(classes[labels[idx]])\n",
    "plt.savefig('plots/data_vis.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 459
    },
    "colab_type": "code",
    "id": "J0yiBCFZQWU5",
    "outputId": "2461da20-79b7-4604-d8d9-f4d448989771"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Net(\n",
       "  (layer1): Sequential(\n",
       "    (0): Conv2d(3, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "    (1): ReLU()\n",
       "    (2): Conv2d(16, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "    (3): ReLU()\n",
       "    (4): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
       "    (5): Conv2d(16, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "    (6): ReLU()\n",
       "    (7): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "    (8): ReLU()\n",
       "    (9): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
       "    (10): Conv2d(32, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "    (11): ReLU()\n",
       "    (12): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
       "  )\n",
       "  (fc): Sequential(\n",
       "    (0): Linear(in_features=4096, out_features=500, bias=True)\n",
       "    (1): ReLU()\n",
       "    (2): Dropout(p=0.2)\n",
       "    (3): Linear(in_features=500, out_features=100, bias=True)\n",
       "    (4): ReLU()\n",
       "    (5): Dropout(p=0.2)\n",
       "    (6): Linear(in_features=100, out_features=2, bias=True)\n",
       "  )\n",
       ")"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# define the CNN architecture\n",
    "class Net(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(Net, self).__init__()\n",
    "        \n",
    "        self.layer1 = nn.Sequential(\n",
    "            nn.Conv2d(3, 16, kernel_size=3, padding=1),\n",
    "            nn.ReLU(),\n",
    "            nn.Conv2d(16, 16, kernel_size=3, padding=1),\n",
    "            nn.ReLU(),\n",
    "            nn.MaxPool2d(2),\n",
    "            nn.Conv2d(16, 32, kernel_size=3, padding=1),\n",
    "            nn.ReLU(),\n",
    "            nn.Conv2d(32, 32, kernel_size=3, padding=1),\n",
    "            nn.ReLU(),\n",
    "            nn.MaxPool2d(2),\n",
    "            nn.Conv2d(32, 64, kernel_size=3, padding=1),\n",
    "            nn.ReLU(),\n",
    "            nn.MaxPool2d(2)\n",
    "        )\n",
    "        self.fc = nn.Sequential(\n",
    "            nn.Linear(8*8*64, 500),\n",
    "            nn.ReLU(),\n",
    "            nn.Dropout(p=0.2),\n",
    "            nn.Linear(500, 100),\n",
    "            nn.ReLU(),\n",
    "            nn.Dropout(p=0.2),\n",
    "            nn.Linear(100, 2)\n",
    "        )\n",
    "    \n",
    "    def forward(self, x):\n",
    "        out = self.layer1(x)\n",
    "        out = out.view(out.shape[0], -1)\n",
    "        out = self.fc(out)\n",
    "        return out\n",
    "\n",
    "\n",
    "model_scratch = Net()\n",
    "model_scratch.to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Ro8Eu6QIUUNP"
   },
   "outputs": [],
   "source": [
    "criterion_scratch = nn.CrossEntropyLoss()\n",
    "\n",
    "# optimizer_scratch = optim.SGD(model_scratch.parameters(), lr=0.001, momentum=0.9)\n",
    "optimizer_scratch = optim.Adam(model_scratch.parameters(), lr=0.001)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "2RoFcWVwTYvt"
   },
   "outputs": [],
   "source": [
    "def train(n_epochs, loaders, model, optimizer, criterion, device, save_path):\n",
    "    \"\"\"returns trained model\"\"\"\n",
    "    # initialize tracker for minimum validation loss\n",
    "    valid_loss_min = np.Inf \n",
    "    train_losses = []\n",
    "    val_losses = []\n",
    "    \n",
    "    for epoch in range(1, n_epochs+1):\n",
    "        # initialize variables to monitor training and validation loss\n",
    "        train_loss = 0.0\n",
    "        valid_loss = 0.0\n",
    "        \n",
    "        ###################\n",
    "        # train the model #\n",
    "        ###################\n",
    "        model.train()\n",
    "        for batch_idx, (data, target) in enumerate(loaders['train']):\n",
    "            # move to GPU or CPU\n",
    "            data, target = data.to(device), target.to(device)\n",
    "            \n",
    "            ## find the loss and update the model parameters accordingly\n",
    "            ## record the average training loss as follows\n",
    "            ## train_loss = train_loss + ((1 / (batch_idx + 1)) * (loss.data - train_loss))\n",
    "            optimizer.zero_grad()\n",
    "            output = model(data)\n",
    "            loss = criterion(output, target)\n",
    "            \n",
    "            loss.backward()\n",
    "            optimizer.step()\n",
    "            \n",
    "            train_loss += ((1 / (batch_idx + 1)) * (loss.data - train_loss))\n",
    "            \n",
    "        ######################    \n",
    "        # validate the model #\n",
    "        ######################\n",
    "        model.eval()\n",
    "        for batch_idx, (data, target) in enumerate(loaders['valid']):\n",
    "            # move to GPU or CPU\n",
    "            data, target = data.to(device), target.to(device)\n",
    "            \n",
    "            ## update the average validation loss\n",
    "            output = model(data)\n",
    "            loss = criterion(output, target)\n",
    "            \n",
    "            valid_loss += ((1 / (batch_idx + 1)) * (loss.data - valid_loss))\n",
    "            \n",
    "        # print training/validation statistics \n",
    "        print('Epoch: {} \\tTraining Loss: {:.6f} \\tValidation Loss: {:.6f}'.format(\n",
    "            epoch, \n",
    "            train_loss,\n",
    "            valid_loss\n",
    "            ))\n",
    "        \n",
    "        train_losses.append(train_loss)\n",
    "        val_losses.append(valid_loss)\n",
    "        \n",
    "        if valid_loss < valid_loss_min:\n",
    "            print('Saving model..')\n",
    "            valid_loss_min = valid_loss\n",
    "            torch.save(model.state_dict(), save_path)\n",
    "    \n",
    "    # return trained model\n",
    "    return model, train_losses, val_losses"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1286
    },
    "colab_type": "code",
    "id": "ToF8RLJtU_LP",
    "outputId": "594a242b-8129-4666-e99b-dc891915597a",
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 1 \tTraining Loss: 0.449945 \tValidation Loss: 0.360567\n",
      "Saving model..\n",
      "Epoch: 2 \tTraining Loss: 0.357576 \tValidation Loss: 0.339078\n",
      "Saving model..\n",
      "Epoch: 3 \tTraining Loss: 0.332712 \tValidation Loss: 0.341607\n",
      "Epoch: 4 \tTraining Loss: 0.328441 \tValidation Loss: 0.318959\n",
      "Saving model..\n",
      "Epoch: 5 \tTraining Loss: 0.318843 \tValidation Loss: 0.338371\n",
      "Epoch: 6 \tTraining Loss: 0.311118 \tValidation Loss: 0.307686\n",
      "Saving model..\n",
      "Epoch: 7 \tTraining Loss: 0.304950 \tValidation Loss: 0.301697\n",
      "Saving model..\n",
      "Epoch: 8 \tTraining Loss: 0.309147 \tValidation Loss: 0.304863\n",
      "Epoch: 9 \tTraining Loss: 0.307798 \tValidation Loss: 0.302695\n",
      "Epoch: 10 \tTraining Loss: 0.299762 \tValidation Loss: 0.323098\n",
      "Epoch: 11 \tTraining Loss: 0.295816 \tValidation Loss: 0.306051\n",
      "Epoch: 12 \tTraining Loss: 0.293260 \tValidation Loss: 0.285249\n",
      "Saving model..\n",
      "Epoch: 13 \tTraining Loss: 0.292022 \tValidation Loss: 0.284853\n",
      "Saving model..\n",
      "Epoch: 14 \tTraining Loss: 0.289929 \tValidation Loss: 0.308076\n",
      "Epoch: 15 \tTraining Loss: 0.294150 \tValidation Loss: 0.297465\n",
      "Epoch: 16 \tTraining Loss: 0.289050 \tValidation Loss: 0.303159\n",
      "Epoch: 17 \tTraining Loss: 0.287979 \tValidation Loss: 0.298820\n",
      "Epoch: 18 \tTraining Loss: 0.287851 \tValidation Loss: 0.311794\n",
      "Epoch: 19 \tTraining Loss: 0.288941 \tValidation Loss: 0.289271\n",
      "Epoch: 20 \tTraining Loss: 0.281930 \tValidation Loss: 0.294516\n"
     ]
    }
   ],
   "source": [
    "# train the model\n",
    "model_scratch, train_losses, val_losses = train(20, dataloaders, model_scratch, optimizer_scratch, \n",
    "                                                criterion_scratch, device, 'models/model_scratch-img64.pt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(train_losses, label='Training loss')\n",
    "plt.plot(val_losses, label='Validation loss')\n",
    "plt.xlabel('Iterations')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "_ = plt.ylim()\n",
    "# plt.title('Training Validation Loss with CNN from scratch')\n",
    "plt.savefig('plots/Training Validation Loss with CNN from scratch.png')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "e_whUDl2VuOi"
   },
   "outputs": [],
   "source": [
    "# load the model that got the best validation accuracy\n",
    "model_scratch.load_state_dict(torch.load('models/model_scratch-img64.pt'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "b66HUWouVydb"
   },
   "outputs": [],
   "source": [
    "def test(loaders, model, criterion, device):\n",
    "    sigmoid = lambda x: 1 / (1 + np.exp(-x))\n",
    "    \n",
    "    # monitor test loss and accuracy\n",
    "    test_loss = 0.\n",
    "    correct = 0.\n",
    "    total = 0.\n",
    "    y = None\n",
    "    y_hat = None\n",
    "    \n",
    "    model.eval()\n",
    "    for batch_idx, (data, target) in enumerate(loaders['test']):\n",
    "        # move to GPU or CPU\n",
    "        data, target = data.to(device), target.to(device)\n",
    "        \n",
    "        # forward pass: compute predicted outputs by passing inputs to the model\n",
    "        output = model(data)\n",
    "        # calculate the loss\n",
    "        loss = criterion(output, target)\n",
    "        # update average test loss \n",
    "        test_loss = test_loss + ((1 / (batch_idx + 1)) * (loss.data - test_loss))\n",
    "        # convert output probabilities to predicted class\n",
    "        pred = output.data.max(1, keepdim=True)[1]\n",
    "        \n",
    "        if y is None:\n",
    "            y = target.cpu().numpy()\n",
    "            y_hat = pred.data.cpu().view_as(target).numpy()\n",
    "        else:\n",
    "            y = np.append(y, target.cpu().numpy())\n",
    "            y_hat = np.append(y_hat, pred.data.cpu().view_as(target).numpy())\n",
    "            \n",
    "        # compare predictions to true label\n",
    "        correct += np.sum(np.squeeze(pred.eq(target.data.view_as(pred))).cpu().numpy())\n",
    "        total += data.size(0)\n",
    "            \n",
    "    print('Test Loss: {:.6f}\\n'.format(test_loss))\n",
    "\n",
    "    print('\\nTest Accuracy: %2d%% (%2d/%2d)' % (\n",
    "        100. * correct / total, correct, total))\n",
    "    \n",
    "    return y, y_hat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 85
    },
    "colab_type": "code",
    "id": "xjPJHHFhn-tT",
    "outputId": "44368847-5be5-403e-e042-29c78e7d6875"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Loss: 0.278942\n",
      "\n",
      "\n",
      "Test Accuracy: 88% (2429/2756)\n"
     ]
    }
   ],
   "source": [
    "# call test function    \n",
    "y, y_hat = test(dataloaders, model_scratch, criterion_scratch, device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "4Uq50WGDgFCy"
   },
   "outputs": [],
   "source": [
    "cm = MetricTools.confusion_matrix(y, y_hat, nclasses=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 297
    },
    "colab_type": "code",
    "id": "34ioObSMqD-3",
    "outputId": "b7452cb1-b9c5-4023-94e4-e010b1db5fd9"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x432 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "PlotTools.confusion_matrix(cm, ['Parasitized', 'Uninfected'], title='',\n",
    "                           filename='Confusion Matrix with CNN from scratch', figsize=(6,6), path='plots/')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "c5PuktzrqGHw"
   },
   "outputs": [],
   "source": [
    "prec, rec = MetricTools.prec_recall(y, y_hat, 2)\n",
    "f1 = 1 /(0.5 * (1/prec + 1/rec))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "7VxDhXHhsqQI",
    "outputId": "713a8a3f-ab88-4399-bd72-da726de1c7dd"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([0.932, 0.828, 0.877]),)"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.around([prec, rec, f1], 3), "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "DHkldJ1FuHIk"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "name": "Malaria-Detection-CNN.ipynb",
   "provenance": [],
   "version": "0.3.2"
  },
  "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}