cs480680_assignment1_W24-Copy1.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "n2dYlHWRAA1a"
   },
   "source": [
    "# W24 CS 480/680 assignment 1 (coding part)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "IuEDzbbiAcXo"
   },
   "source": [
    "- Please save a copy of this notebook to avoid losing your changes.\n",
    "- Debug your code and ensure that it can run before submission.\n",
    "- Save the output of each cell. Failure to do so may result in your coding questions not being graded.\n",
    "- Submit your completed version of this notebook.\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "hu7upEwVALqo"
   },
   "source": [
    "## Implement Your Own Perceptron\n",
    "\n",
    "In this question, you are asked to implement the **Perceptron** algorithm and test with a classification task on the adult dataset. Please refer to the [dataset webpage](https://archive.ics.uci.edu/dataset/53/iris) for details on the dataset, and consult the lecture slides and suggested readings for details on the Perceptron algorithm.\n",
    "\n",
    "- `scikit-learn` or other off-the-shelf Perceptron libraries are not allowed to use.\n",
    "\n",
    "- Please note that `class` column is the label of the dataset. The labellings have three possible values. So you need to implement a multi-class extension of Perceptron.\n",
    "\n",
    "- Recording the accuracy on the training dataset when you train the Perceptron model on the training data, as an iteration. Repeat this step for 10 times.\n",
    "\n",
    "- It is required that you split the dataset into training/testing datasets (80% and 20%, respectively), and we will deduct marks if you don't.\n",
    "\n",
    "- You will get full marks for\n",
    "    - Correctly split data into training set and testing set.\n",
    "    - Correctly implement the perceptron algorithm.\n",
    "    - Correctly handle multiple class.\n",
    "    - A plot on training accuracy vs number of iterations, that relects the progress of the training.\n",
    "    - Final reported accuracy (on the test dataset)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "id": "gl0lxf-V-LYF"
   },
   "outputs": [],
   "source": [
    "import urllib.request\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "# TODO: add any other package you need\n",
    "\n",
    "# Download the dataset\n",
    "url = 'https://archive.ics.uci.edu/ml/machine-learning-databases/iris/iris.data'\n",
    "filename = 'iris.csv'\n",
    "urllib.request.urlretrieve(url, filename)\n",
    "\n",
    "column_names = [\n",
    "    'sepal_length', 'sepal_width', 'petal_length', 'petal_width', 'class'\n",
    "]\n",
    "data = pd.read_csv(filename, names=column_names)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 206
    },
    "id": "LUWHsWBCiq2f",
    "outputId": "4929de01-c7aa-4ad8-96ab-1ccecb0a9aa2"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>sepal_length</th>\n",
       "      <th>sepal_width</th>\n",
       "      <th>petal_length</th>\n",
       "      <th>petal_width</th>\n",
       "      <th>class</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.1</td>\n",
       "      <td>3.5</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>Iris-setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4.9</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>Iris-setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>4.7</td>\n",
       "      <td>3.2</td>\n",
       "      <td>1.3</td>\n",
       "      <td>0.2</td>\n",
       "      <td>Iris-setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4.6</td>\n",
       "      <td>3.1</td>\n",
       "      <td>1.5</td>\n",
       "      <td>0.2</td>\n",
       "      <td>Iris-setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5.0</td>\n",
       "      <td>3.6</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>Iris-setosa</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   sepal_length  sepal_width  petal_length  petal_width        class\n",
       "0           5.1          3.5           1.4          0.2  Iris-setosa\n",
       "1           4.9          3.0           1.4          0.2  Iris-setosa\n",
       "2           4.7          3.2           1.3          0.2  Iris-setosa\n",
       "3           4.6          3.1           1.5          0.2  Iris-setosa\n",
       "4           5.0          3.6           1.4          0.2  Iris-setosa"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "_covQQeGU0bK"
   },
   "source": [
    "Code for dataset splitting (training-testing):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "id": "xJ6Xw9T-CHQg"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "30\n"
     ]
    }
   ],
   "source": [
    "# TODO: your implementation\n",
    "X = data.drop('class', axis=1)\n",
    "y = data['class']\n",
    "n = X.shape[0]\n",
    "indices = np.arange(n)\n",
    "np.random.shuffle(indices)\n",
    "X = X.iloc[indices]\n",
    "y = y.iloc[indices]\n",
    "    \n",
    "train = int(n * 0.8)\n",
    "X_train = X.iloc[:train]\n",
    "y_train = y.iloc[:train]\n",
    "X_test = X.iloc[train:]\n",
    "y_test = y.iloc[train:]\n",
    "\n",
    "print(n - train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ds-7f6JzVDq3"
   },
   "source": [
    "Code for implemeting the Perceptron algorithm:\n",
    "- You can either implement it as an class/object or implement the training function and the prediction function."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "id": "FIq0ss_DVAnU"
   },
   "outputs": [],
   "source": [
    "# TODO: your implementation\n",
    "class Perceptron:\n",
    "    def __init__(self, learning_rate, iterations):\n",
    "        self.lr = learning_rate\n",
    "        self.iterations = iterations\n",
    "        self.weights = None\n",
    "        self.bias = None\n",
    "\n",
    "    def train(self, X, y):\n",
    "        n_samples, n_features = X.shape\n",
    "        self.weights = np.zeros(n_features)\n",
    "        self.bias = 0\n",
    "        self.errors = []\n",
    "\n",
    "        for _ in range(self.iterations):\n",
    "            errors = 0\n",
    "            for idx, x_i in enumerate(X):\n",
    "                y_cur = np.dot(x_i, self.weights) + self.bias\n",
    "                if y_cur * y[idx] <= 0:\n",
    "                    update = self.lr * y[idx]\n",
    "                    self.weights += update * x_i\n",
    "                    self.bias += update\n",
    "                    errors += 1\n",
    "            self.errors.append(errors)\n",
    "        return self\n",
    "\n",
    "\n",
    "    def predict(self, X):\n",
    "        linear_output = np.dot(X, self.weights) + self.bias\n",
    "        return linear_output\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Rp0H9r1VYTwg"
   },
   "source": [
    "Code for implementing a wrapper function:\n",
    "- Handling multi-class extension\n",
    "- Feeding in data and call Perceptron training function\n",
    "- Outputing multiple Perceptrons that are trained"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "id": "jXLprdF8YMyt"
   },
   "outputs": [],
   "source": [
    "# TODO: your implementation\n",
    "# One vs. All\n",
    "def train_perceptrons(X, y):\n",
    "    classes = np.unique(y)\n",
    "    perceptrons = {}\n",
    "\n",
    "    for cls in classes:\n",
    "        y_binary = np.where(y == cls, 1, -1)\n",
    "        perceptron = Perceptron(1, 10)\n",
    "        perceptron.train(X, y_binary)\n",
    "        perceptrons[cls] = perceptron\n",
    "\n",
    "    return perceptrons\n",
    "\n",
    "def predict_multi_perceptron(X, perceptrons):\n",
    "    predictions = {cls: p.predict(X.values) for cls, p in perceptrons.items()}\n",
    "    predictions_df = pd.DataFrame(predictions)\n",
    "    ## print(predictions_df)\n",
    "    \n",
    "    predicted_classes = predictions_df.idxmax(axis=1)\n",
    "    return predicted_classes\n",
    "\n",
    "# Training Perceptrons\n",
    "perceptrons = train_perceptrons(X_train.values, y_train.values)\n",
    "\n",
    "predicted_classes = predict_multi_perceptron(X_test, perceptrons)\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "hqC_G4RGVKzr"
   },
   "source": [
    "Code for testing and ploting:\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "id": "hy2lir66C2AJ"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# TODO: plot the accuracy against the number of iterations\n",
    "def plot_perceptron_training(perceptrons, X):\n",
    "    n_samples = X.shape[0] \n",
    "\n",
    "    for cls, perceptron in perceptrons.items():\n",
    "        accuracies = [(1 - (error / n_samples)) for error in perceptron.errors]\n",
    "        plt.plot(accuracies, label=f'Class {cls}')\n",
    "\n",
    "    plt.xlabel('Iterations')\n",
    "    plt.ylabel('Accuracy')\n",
    "    plt.legend()\n",
    "    plt.title('Perceptron Training Accuracy')\n",
    "    plt.show()\n",
    "\n",
    "plot_perceptron_training(perceptrons, X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Overall Accuracy: 76.67%\n"
     ]
    }
   ],
   "source": [
    "# TODO: report final accuracy\n",
    "def test_perceptrons(X, y, perceptrons):\n",
    "    predicted_classes = predict_multi_perceptron(X, perceptrons)\n",
    "    ## print(predicted_classes.values == y.values)\n",
    "    accuracy = np.mean(predicted_classes.values == y.values)\n",
    "    print(f\"Overall Accuracy: {accuracy * 100:.2f}%\")\n",
    "\n",
    "test_perceptrons(X_test, y_test, perceptrons)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "provenance": []
  },
  "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": "3.9.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}