kfold-test.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/foxtrotmike/CS909/blob/master/kfold-test.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 60,
      "metadata": {
        "id": "bP-anSFWtxMC"
      },
      "outputs": [],
      "source": [
        "import numpy as np\n",
        "from sklearn.model_selection import KFold\n",
        "from sklearn.linear_model import LogisticRegression\n",
        "from sklearn.metrics import roc_auc_score\n",
        "import matplotlib.pyplot as plt\n",
        "from numpy.random import randn\n",
        "import numpy as np\n",
        "\n",
        "class DataStreamGenerator:\n",
        "    \"\"\"\n",
        "    A data generator class that produces one example at a time\n",
        "    with a fixed proportion of positive and negative examples.\n",
        "    \"\"\"\n",
        "    def __init__(self, pos_ratio=0.5, d=2, random_seed=None):\n",
        "        \"\"\"\n",
        "        Initialize the data stream generator.\n",
        "\n",
        "        Parameters:\n",
        "        - pos_ratio (float): Proportion of positive examples (0.0 to 1.0).\n",
        "        - d (int): Dimensionality of the data (number of features).\n",
        "        - random_seed (int or None): Random seed for reproducibility.\n",
        "        \"\"\"\n",
        "        assert 0 <= pos_ratio <= 1, \"pos_ratio must be between 0 and 1.\"\n",
        "        self.pos_ratio = pos_ratio\n",
        "        self.d = d\n",
        "        self.random_seed = random_seed\n",
        "        if random_seed is not None:\n",
        "            np.random.seed(random_seed)\n",
        "\n",
        "    def generate_example(self):\n",
        "        \"\"\"\n",
        "        Generate a single example (one data point).\n",
        "\n",
        "        Returns:\n",
        "        - x (numpy.ndarray): Feature vector of shape (d,).\n",
        "        - y (int): Label (+1 for positive, -1 for negative).\n",
        "        \"\"\"\n",
        "        if np.random.rand() < self.pos_ratio:\n",
        "            # Generate a positive example\n",
        "            x = randn(self.d) + 1\n",
        "            y = +1\n",
        "        else:\n",
        "            # Generate a negative example from one of two clusters\n",
        "            if np.random.rand() < 0.5:\n",
        "                x = randn(self.d) - 1  # Cluster 1\n",
        "            else:\n",
        "                x = randn(self.d) + 2  # Cluster 2\n",
        "            y = -1\n",
        "        return x, y\n",
        "\n",
        "    def generate_batch(self, n):\n",
        "        \"\"\"\n",
        "        Generate a batch of examples.\n",
        "\n",
        "        Parameters:\n",
        "        - n (int): Number of examples to generate.\n",
        "\n",
        "        Returns:\n",
        "        - X (numpy.ndarray): Feature matrix of shape (n, d).\n",
        "        - Y (numpy.ndarray): Labels array of shape (n,).\n",
        "        \"\"\"\n",
        "        X, Y = [], []\n",
        "        for _ in range(n):\n",
        "            x, y = self.generate_example()\n",
        "            X.append(x)\n",
        "            Y.append(y)\n",
        "        return np.array(X), np.array(Y)\n",
        "\n",
        "    def __repr__(self):\n",
        "        \"\"\"\n",
        "        Representation of the DataStreamGenerator object.\n",
        "        \"\"\"\n",
        "        return (f\"DataStreamGenerator(pos_ratio={self.pos_ratio}, \"\n",
        "                f\"d={self.d}, random_seed={self.random_seed})\")\n",
        "\n",
        "\n",
        "import numpy as np\n",
        "from sklearn.model_selection import StratifiedKFold\n",
        "from sklearn.svm import SVC\n",
        "from sklearn.metrics import roc_auc_score\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "# Initialize the data generator\n",
        "generator = DataStreamGenerator(pos_ratio=0.2, d=2, random_seed=42)\n",
        "\n",
        "# Generate training and testing data\n",
        "X_train, y_train = generator.generate_batch(200)\n",
        "\n",
        "# Function to evaluate cross-validation accuracy and AUROC with skipping and stratified KFold\n",
        "def evaluate_kfold_auroc_svm(X_train, y_train, max_k=10, skip=2, num_runs=5):\n",
        "    all_fold_aurocs = {}\n",
        "\n",
        "    for k in range(2, max_k + 1, skip):  # Skip every other K based on the `skip` parameter\n",
        "        kfold_aurocs = []\n",
        "\n",
        "        for run in range(num_runs):  # Perform multiple runs\n",
        "            skf = StratifiedKFold(n_splits=k, shuffle=True, random_state=42 + run)\n",
        "\n",
        "            fold_aurocs = []\n",
        "\n",
        "            # Perform stratified K-fold cross-validation\n",
        "            for train_idx, val_idx in skf.split(X_train, y_train):\n",
        "                X_train_fold, X_val_fold = X_train[train_idx], X_train[val_idx]\n",
        "                y_train_fold, y_val_fold = y_train[train_idx], y_train[val_idx]\n",
        "\n",
        "                # Train an SVM model\n",
        "                model = SVC(C=10.0, kernel='rbf', gamma=1.0, probability=True)\n",
        "                model.fit(X_train_fold, y_train_fold)\n",
        "\n",
        "                # Predict probabilities for validation fold\n",
        "                y_val_pred = model.predict_proba(X_val_fold)[:, 1]\n",
        "                fold_aurocs.append(roc_auc_score(y_val_fold, y_val_pred))\n",
        "\n",
        "            kfold_aurocs.extend(fold_aurocs)\n",
        "\n",
        "        all_fold_aurocs[k] = kfold_aurocs\n",
        "\n",
        "    return all_fold_aurocs\n",
        "\n",
        "# Function to plot the box plot and compute AUROC on the test set\n",
        "def plot_kfold_aurocs(all_fold_aurocs):\n",
        "    ks = sorted(all_fold_aurocs.keys())\n",
        "    data = [all_fold_aurocs[k] for k in ks]\n",
        "    median_aurocs = [np.nanmedian(all_fold_aurocs[k]) for k in ks]\n",
        "\n",
        "    # Box plot\n",
        "    plt.figure(figsize=(10, 6))\n",
        "    plt.boxplot(data, labels=ks, showmeans=True)\n",
        "    plt.plot(range(1, len(ks) + 1), median_aurocs, marker='o', linestyle='-', color='red', label='Mean AUROC')\n",
        "    plt.title('Cross-Validation AUROC at Different K Values (Stratified K-Fold)')\n",
        "    plt.xlabel('Number of Folds (K)')\n",
        "    plt.ylabel('Validation AUROC')\n",
        "    plt.legend()\n",
        "    plt.grid(True)\n",
        "    plt.show()\n",
        "\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Evaluate AUROC for K values and plot\n",
        "max_k = 50  # Maximum K to test\n",
        "skip = 2    # Skip every other K\n",
        "num_runs = 20  # Number of runs\n",
        "all_fold_aurocs = evaluate_kfold_auroc_svm(X_train, y_train, max_k=max_k, skip=skip, num_runs=num_runs)\n"
      ],
      "metadata": {
        "id": "VajZSJ_LukdV"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plot_kfold_aurocs(all_fold_aurocs)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 619
        },
        "id": "R-fuAbi1ulz0",
        "outputId": "87a47a92-23d8-4dc6-e539-c33eddb733f5"
      },
      "execution_count": 61,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "<ipython-input-60-9cd67a118e0f>:128: MatplotlibDeprecationWarning: The 'labels' parameter of boxplot() has been renamed 'tick_labels' since Matplotlib 3.9; support for the old name will be dropped in 3.11.\n",
            "  plt.boxplot(data, labels=ks, showmeans=True)\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1000x600 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 62,
      "metadata": {
        "id": "uAHLCxpKtyFl",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "4dfbdbca-ce33-4032-b8c8-74489c12b74c"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Average AUROC for k=2: 0.6369\n",
            "Average AUROC for k=4: 0.6759\n",
            "Average AUROC for k=6: 0.6568\n",
            "Average AUROC for k=8: 0.6667\n",
            "Average AUROC for k=10: 0.6982\n",
            "Average AUROC for k=12: 0.6923\n",
            "Average AUROC for k=14: 0.6970\n",
            "Average AUROC for k=16: 0.7000\n",
            "Average AUROC for k=18: 0.7083\n",
            "Average AUROC for k=20: 0.6875\n",
            "Average AUROC for k=22: 0.6429\n",
            "Average AUROC for k=24: 0.7143\n",
            "Average AUROC for k=26: 0.6833\n",
            "Average AUROC for k=28: 0.7000\n",
            "Average AUROC for k=30: 0.7000\n",
            "Average AUROC for k=32: 0.7500\n",
            "Average AUROC for k=34: 0.7500\n",
            "Average AUROC for k=36: 0.7500\n",
            "Average AUROC for k=38: 0.7500\n",
            "Average AUROC for k=40: 0.7500\n",
            "Average AUROC for k=42: 0.7500\n",
            "Average AUROC for k=44: 0.7500\n",
            "Average AUROC for k=46: 0.7500\n",
            "Average AUROC for k=48: 0.7500\n",
            "Average AUROC for k=50: 0.6667\n"
          ]
        }
      ],
      "source": [
        "import numpy as np\n",
        "average_aurocs = {}\n",
        "for k, aurocs in all_fold_aurocs.items():\n",
        "    average_aurocs[k] = np.nanmedian(aurocs)\n",
        "\n",
        "# Print the results\n",
        "for k, avg_auroc in average_aurocs.items():\n",
        "    print(f\"Average AUROC for k={k}: {avg_auroc:.4f}\")"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Train the final model\n",
        "final_model = SVC(C=10.0, kernel='rbf', gamma=1.0, probability=True)\n",
        "final_model.fit(X_train, y_train)\n",
        "test_aurocs = []\n",
        "test_size = 2000\n",
        "num_runs = 20\n",
        "for _ in range(num_runs):\n",
        "    # Generate a new test set for each run\n",
        "    X_test, y_test = generator.generate_batch(test_size)\n",
        "    # Predict decision function and calculate AUROC\n",
        "    y_test_pred = final_model.decision_function(X_test)\n",
        "    test_auroc = roc_auc_score(y_test, y_test_pred)\n",
        "    test_aurocs.append(test_auroc)\n",
        "\n",
        "# Calculate mean and standard deviation\n",
        "mean_auroc = np.mean(test_aurocs)\n",
        "std_auroc = np.std(test_aurocs)\n",
        "\n",
        "print(f\"Test AUROC (Mean over {num_runs} runs): {mean_auroc:.4f}\")\n",
        "print(f\"Test AUROC (Standard Deviation over {num_runs} runs): {std_auroc:.4f}\")\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BPv0UnwanL0W",
        "outputId": "b93bb378-1fe6-4ca7-b155-a0b9bab0dae0"
      },
      "execution_count": 65,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Test AUROC (Mean over 20 runs): 0.7562\n",
            "Test AUROC (Standard Deviation over 20 runs): 0.0113\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.utils import resample\n",
        "from sklearn.svm import SVC\n",
        "from sklearn.metrics import roc_auc_score\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "def stratified_bootstrap_auroc(X_train, y_train, num_iterations=100):\n",
        "    \"\"\"\n",
        "    Perform stratified bootstrap to estimate AUROC on out-of-sample data.\n",
        "\n",
        "    Parameters:\n",
        "    - X_train (np.ndarray): Training features.\n",
        "    - y_train (np.ndarray): Training labels.\n",
        "    - num_iterations (int): Number of bootstrap iterations.\n",
        "\n",
        "    Returns:\n",
        "    - auroc_scores (list): List of AUROC scores from each bootstrap iteration.\n",
        "    \"\"\"\n",
        "    auroc_scores = []\n",
        "\n",
        "    for i in range(num_iterations):\n",
        "        # Perform stratified bootstrap resampling\n",
        "        indices = np.arange(len(y_train))\n",
        "        resampled_indices = resample(indices, stratify=y_train, random_state=42 + i)\n",
        "\n",
        "        # Identify the out-of-sample indices\n",
        "        out_of_sample_mask = np.ones(len(y_train), dtype=bool)\n",
        "        out_of_sample_mask[resampled_indices] = False\n",
        "        out_of_sample_indices = np.where(out_of_sample_mask)[0]\n",
        "\n",
        "        # Skip this iteration if the out-of-sample set is empty\n",
        "        if len(out_of_sample_indices) == 0:\n",
        "            continue\n",
        "\n",
        "        X_resampled = X_train[resampled_indices]\n",
        "        y_resampled = y_train[resampled_indices]\n",
        "        X_out_of_sample = X_train[out_of_sample_indices]\n",
        "        y_out_of_sample = y_train[out_of_sample_indices]\n",
        "\n",
        "        # Train an SVM model on the resampled data\n",
        "        model = SVC(C=10.0, kernel='rbf', gamma=1.0, probability=True)\n",
        "        model.fit(X_resampled, y_resampled)\n",
        "\n",
        "        # Evaluate AUROC on the out-of-sample data\n",
        "        y_pred_out_of_sample = model.predict_proba(X_out_of_sample)[:, 1]\n",
        "        auroc = roc_auc_score(y_out_of_sample, y_pred_out_of_sample)\n",
        "        auroc_scores.append(auroc)\n",
        "\n",
        "    return auroc_scores\n",
        "\n",
        "# Run stratified bootstrap to estimate AUROC\n",
        "num_iterations = 100  # Number of bootstrap iterations\n",
        "bootstrap_aurocs = stratified_bootstrap_auroc(X_train, y_train, num_iterations=num_iterations)\n",
        "\n",
        "# Print and plot the results\n",
        "print(f\"Bootstrap AUROC (Mean): {np.mean(bootstrap_aurocs):.4f}\")\n",
        "print(f\"Bootstrap AUROC (Std): {np.std(bootstrap_aurocs):.4f}\")\n",
        "\n",
        "plt.figure(figsize=(8, 6))\n",
        "plt.hist(bootstrap_aurocs, bins=15, edgecolor='k', alpha=0.7)\n",
        "plt.title(\"Distribution of Out-of-Sample AUROC (Stratified Bootstrap)\", fontsize=14)\n",
        "plt.xlabel(\"AUROC\", fontsize=12)\n",
        "plt.ylabel(\"Frequency\", fontsize=12)\n",
        "plt.grid(True)\n",
        "plt.show()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 604
        },
        "id": "9sDUsixWncRl",
        "outputId": "760eb03f-0ddb-4602-c00b-eb4cf6be7777"
      },
      "execution_count": 57,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Bootstrap AUROC (Mean): 0.6520\n",
            "Bootstrap AUROC (Std): 0.0597\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 800x600 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Calculate the two-tailed p-value\n",
        "p_left = np.sum(np.array(bootstrap_aurocs) <= test_auroc) / len(bootstrap_aurocs)\n",
        "p_right = np.sum(np.array(bootstrap_aurocs) >= test_auroc) / len(bootstrap_aurocs)\n",
        "p_value = 2 * min(p_left, p_right)\n",
        "\n",
        "print(f\"Two-tailed p-value: {p_value:.4f}\")\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "dlCURb-upmxv",
        "outputId": "dbdf5b77-6736-4836-bd87-8f85e90c2f18"
      },
      "execution_count": 58,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Two-tailed p-value: 0.0800\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "o5Rqeq_sqL7q"
      },
      "execution_count": null,
      "outputs": []
    }
  ],
  "metadata": {
    "colab": {
      "provenance": [],
      "authorship_tag": "ABX9TyOWW+N5ug/QN/qonclNWga1",
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}