{ "cells": [ { "cell_type": "code", "execution_count": 1, "id": "ae94a946-086b-41b9-bcc6-7750b96ba59e", "metadata": {}, "outputs": [], "source": [ "import torchvision #! pip3 install torchvision\n", "import torchvision.transforms as transforms\n", "from torchvision.datasets import MNIST\n", "import matplotlib.pyplot as plt\n", "\n", "transform = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.5,), (1.0,))])\n", "trainset = MNIST(root = './', train=True, download=True, transform=transform)\n", "testset = MNIST(root = './', train=False, download=True, transform=transform)" ] }, { "cell_type": "code", "execution_count": 2, "id": "62a1c998-9dcd-466e-b0c6-8faf9476b307", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Number of observations in train set: 60000\n", " Number of observations in test set: 10000\n" ] }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "print(f\"Number of observations in train set: {trainset.data.shape[0]}\")\n", "print(f\" Number of observations in test set: {testset.data.shape[0]}\")\n", "\n", "# plotting an image in the dataset using imshow()\n", "idx = 2489\n", "plt.imshow(trainset.data[idx], cmap='gray')\n", "plt.title(trainset.targets[idx].item())\n", "plt.show()" ] }, { "cell_type": "markdown", "id": "30b6e8bd-2a4f-4a42-af9f-16f837c11b52", "metadata": {}, "source": [ "# Binary" ] }, { "cell_type": "code", "execution_count": 11, "id": "4f00b5a9-0000-4b38-a394-264b3f9b9ab8", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Xtrain shape = (1200, 10)\n", "Xvalid shape = (400, 10)\n", "Xtest shape = (400, 10)\n" ] } ], "source": [ "from sklearn.model_selection import train_test_split\n", "import numpy as np\n", "\n", "# sample size and features to work with\n", "s = 10 # reduced features to work with \n", "sFeatures = np.random.choice(np.arange(784), size=s, replace=False) # choose s features randomly from the 784\n", "sFeatures.sort()\n", "nSample = 1000\n", "\n", "# choose two class labels\n", "class1Label = 4\n", "class2Label = 9\n", "\n", "class1Images = trainset.data[trainset.targets==class1Label].reshape(-1,784).numpy() # images with class1\n", "class2Images = trainset.data[trainset.targets==class2Label].reshape(-1,784).numpy() # images with class2\n", "\n", "# work with a smaller sample size\n", "class1Images = class1Images[:nSample, sFeatures]\n", "class2Images = class2Images[:nSample, sFeatures]\n", "X = np.concatenate((class1Images, class2Images), axis=0)\n", "y = np.concatenate((np.zeros(class1Images.shape[0]), np.ones(class2Images.shape[0])))\n", "\n", "# create Xtrain, Xvalid, Xtest\n", "Xtrain, X_, ytrain, y_ = train_test_split(X, y, test_size=0.4, shuffle=True)\n", "Xvalid, Xtest, yvalid, ytest = train_test_split(X_, y_, test_size=0.5, shuffle=True)\n", "\n", "print(f'Xtrain shape = {Xtrain.shape}')\n", "print(f'Xvalid shape = {Xvalid.shape}')\n", "print(f'Xtest shape = {Xtest.shape}')" ] }, { "cell_type": "code", "execution_count": 12, "id": "1dfc7d52-35de-4077-b568-556cf62e8ad2", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Best C according to gridsearch: 0.002119191919191919\n", "Train Accuracy: 0.48583333333333334\n", "Test Accuracy: 0.5075\n" ] } ], "source": [ "from sklearn.linear_model import LogisticRegression\n", "from sklearn.model_selection import GridSearchCV\n", "from sklearn.metrics import accuracy_score\n", "from sklearn.multiclass import OneVsRestClassifier\n", "import numpy as np\n", "\n", "binary_classes = [0, 1]\n", "\n", "# Filter training data\n", "train_mask = np.isin(ytrain, binary_classes)\n", "Xtrain_bin = Xtrain[train_mask]\n", "ytrain_bin = ytrain[train_mask]\n", "\n", "# Filter test data\n", "test_mask = np.isin(ytest, binary_classes)\n", "Xtest_bin = Xtest[test_mask]\n", "ytest_bin = ytest[test_mask]\n", "\n", "# Filter validation data\n", "valid_mask = np.isin(yvalid, binary_classes)\n", "Xvalid_bin = Xvalid[valid_mask]\n", "yvalid_bin = yvalid[valid_mask]\n", "\n", "C_grid = np.linspace(0.0001, 0.2, 100)\n", "\n", "# Use correct param name for nested estimator\n", "param_grid = {'estimator__C': C_grid}\n", "\n", "grid_lr = GridSearchCV(\n", " estimator=OneVsRestClassifier(LogisticRegression(penalty='l1', solver='liblinear')),\n", " cv=10,\n", " param_grid=param_grid,\n", " scoring='neg_log_loss'\n", ")\n", "grid_lr.fit(Xvalid_bin, yvalid_bin)\n", "Cbest = grid_lr.best_params_['estimator__C']\n", "print(f'Best C according to gridsearch: {Cbest}')\n", "\n", "# Train final model\n", "logistic_mod = OneVsRestClassifier(\n", " LogisticRegression(penalty='l1', solver='liblinear', C=Cbest)\n", ").fit(Xtrain_bin, ytrain_bin)\n", "\n", "print(f'Train Accuracy: {accuracy_score(logistic_mod.predict(Xtrain_bin), ytrain_bin)}')\n", "print(f'Test Accuracy: {accuracy_score(logistic_mod.predict(Xtest_bin), ytest_bin)}')\n" ] }, { "cell_type": "markdown", "id": "d95d9789-1664-44d6-a4b5-69e9e6fd25b2", "metadata": {}, "source": [ "# Metrics" ] }, { "cell_type": "markdown", "id": "74ec1337-efd0-4874-a355-d287bdb4294d", "metadata": {}, "source": [ "## TPR, TNR\n", "- True Positive Rate (TPR), Sensitivity, Recall: $\\text{TPR} = \\frac{\\text{TP}}{\\text{P}} = \\frac{\\text{TP}}{\\text{TP} + \\text{FN}}$ where $P$ is the actual number of positive values in the data.\n", "- True Negative Rate (TNR), Specificity, Selectivity: $\\text{TNR} = \\frac{\\text{TN}}{\\text{N}} = \\frac{\\text{TN}}{\\text{TN} + \\text{FP}}$ where $N$ is the actual number of negative values in the data.\n", "\n", "Any good classifier should ideally have a large TPR and a large TNR, but there is a trade-off between the two." ] }, { "cell_type": "markdown", "id": "919fc12e-c322-4f95-a92a-c3d23f8174bf", "metadata": {}, "source": [ "## ROC\n", "Receiver Operator Characteristic (ROC) curve is a graphical depiction of what happens to the TPR and FPR = 1-TNR as we vary the threshold $t$. It is a good way of comparing different classification models. \n", "\n", "## AUC\n", "The Area Under the Curve (AUC) gives us a nice summary of how good a particular model is, it is the area under each of the ROC curves." ] }, { "cell_type": "code", "execution_count": 13, "id": "660ab45b-eb62-47a5-b635-f51133c3337c", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "from sklearn.metrics import roc_curve, roc_auc_score\n", "\n", "# choosing C via cross validation then computing FPR and TPR at all thresholds\n", "logistic_mod_1 = OneVsRestClassifier(LogisticRegression( penalty='l1', solver='liblinear', C=Cbest)).fit(Xtrain, ytrain)\n", "mod_1_preds = logistic_mod_1.predict_proba(Xtest)[:,1] # note, predictions need to be just for one of the classes\n", "fpr_1, tpr_1, _ = roc_curve(ytest, mod_1_preds)\n", "auc_1 = roc_auc_score(ytest, mod_1_preds)\n", "\n", "# choosing C small\n", "logistic_mod_2 = OneVsRestClassifier(LogisticRegression(penalty='l1', solver='liblinear', C=0.02)).fit(Xtrain, ytrain)\n", "mod_2_preds = logistic_mod_2.predict_proba(Xtest)[:,1]\n", "fpr_2, tpr_2, _ = roc_curve(ytest, mod_2_preds)\n", "auc_2 = roc_auc_score(ytest, mod_2_preds)\n", "\n", "# choosing C to be very large (no regularization)\n", "logistic_mod_3 = OneVsRestClassifier(LogisticRegression(solver='liblinear', C=1000)).fit(Xtrain, ytrain)\n", "mod_3_preds = logistic_mod_3.predict_proba(Xtest)[:,1]\n", "fpr_3, tpr_3, _ = roc_curve(ytest, mod_3_preds)\n", "auc_3 = roc_auc_score(ytest, mod_3_preds)\n", "\n", "\n", "fig = plt.figure(figsize=(10,10))\n", "plt.plot(fpr_1, tpr_1, label=\"model 1, AUC=\"+str(auc_1))\n", "plt.plot(fpr_2, tpr_2, label=\"model 2, AUC=\"+str(auc_2))\n", "plt.plot(fpr_3, tpr_3, label=\"model 3, AUC=\"+str(auc_3))\n", "plt.ylabel(\"True Positive Rate (TPR)\")\n", "plt.xlabel(\"False Positive Rate (FPR)\")\n", "plt.title(\"ROC Curves for Three Logistic Models\")\n", "plt.legend()\n", "plt.show()\n" ] }, { "cell_type": "markdown", "id": "a6094401-fc27-41c5-898e-a9219baffc3c", "metadata": {}, "source": [ "# Multiclass" ] }, { "cell_type": "code", "execution_count": 14, "id": "d41bd8f6-f6e7-4f66-8569-f4251355f4b4", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Xtrain shape = (2000, 784)\n", "Xtest shape = (400, 784)\n" ] } ], "source": [ "# subsample the data to make it more manageable\n", "nSample = 2000\n", "idxsTrain = np.random.choice(np.arange(trainset.data.shape[0]), size=nSample, replace=False)\n", "idxsTest = np.random.choice(np.arange(testset.data.shape[0]), size=nSample//5, replace=False)\n", "\n", "Xtrain = trainset.data.reshape(-1, 784).numpy()[idxsTrain, :]\n", "Xtest = testset.data.reshape(-1, 784).numpy()[idxsTest, :] \n", "ytrain = trainset.targets.numpy()[idxsTrain]\n", "ytest = testset.targets.numpy()[idxsTest]\n", "\n", "print(f'Xtrain shape = {Xtrain.shape}')\n", "print(f'Xtest shape = {Xtest.shape}')" ] }, { "cell_type": "code", "execution_count": 15, "id": "2c25075d-b4b9-47c3-b080-5cc30691300f", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Confusion_matrix: \n", "[[30 0 0 0 0 0 1 0 0 0]\n", " [ 0 37 0 0 0 0 0 0 0 0]\n", " [ 0 0 41 1 1 1 0 0 0 0]\n", " [ 0 1 0 39 0 1 0 1 2 1]\n", " [ 0 0 1 0 35 0 2 0 2 3]\n", " [ 1 0 0 5 2 28 0 0 2 0]\n", " [ 1 0 3 0 1 2 41 0 0 0]\n", " [ 0 2 1 1 0 0 0 32 0 2]\n", " [ 0 1 0 3 2 2 2 0 32 2]\n", " [ 0 0 1 1 3 1 0 1 1 24]]\n" ] } ], "source": [ "from sklearn.metrics import confusion_matrix\n", "\n", "sm_mod = LogisticRegression(#automatically uses `multi_class = 'multinomial'` after scikit v1.5\n", " penalty='l2',\n", " C=50,\n", " solver='sag',\n", " tol=.001,\n", " max_iter=1000\n", " ).fit(Xtrain, ytrain)\n", "\n", "ypred = sm_mod.predict(Xtest)\n", "print(\"Confusion_matrix: \\n\"+str(confusion_matrix(ytest, ypred)))" ] }, { "cell_type": "code", "execution_count": 16, "id": "a1a1952c-2495-4b23-88bd-6b92271b27ed", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "fig, axes = plt.subplots(2, 5, figsize=(14,8))\n", "for i, ax in enumerate(axes.flat):\n", " ax.imshow(sm_mod.coef_[i].reshape(28,28), cmap=plt.cm.RdBu_r, interpolation='nearest')\n", "plt.tight_layout()\n", "plt.show()" ] }, { "cell_type": "code", "execution_count": null, "id": "be0fa796-1bf4-4fb7-bb7d-e7b410450377", "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "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": "3.12.9" } }, "nbformat": 4, "nbformat_minor": 5 }