Iris.ipynb

{
 "cells": [
  {
   "cell_type": "raw",
   "id": "d2d997b1-8174-446a-bc24-6030bed44414",
   "metadata": {},
   "source": [
    "3# Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "30720b62-ffe0-4a32-9c3e-caf74a45803a",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.datasets import load_iris\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "import torch\n",
    "import torch.nn.functional as F\n",
    "import torch.nn as nn\n",
    "from torch.autograd import Variable\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "6eccd85f-f5e2-4a75-bca3-f5fb035263c3",
   "metadata": {},
   "outputs": [],
   "source": [
    "class NeuralNetworkClassificationModel(nn.Module):\n",
    "    def __init__(self,input_dim,output_dim):\n",
    "        super(NeuralNetworkClassificationModel,self).__init__()\n",
    "        self.input_layer    = nn.Linear(input_dim,4)\n",
    "        self.hidden_layer1  = nn.Linear(4,4)\n",
    "        self.output_layer   = nn.Linear(4,output_dim)\n",
    "        self.relu = nn.ReLU()\n",
    "    \n",
    "    \n",
    "    def forward(self,x):\n",
    "        out =  self.relu(self.input_layer(x))\n",
    "        out =  self.relu(self.hidden_layer1(out))\n",
    "        out =  self.output_layer(out)\n",
    "        return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "377d6acc-102d-4d06-8c42-4da0482e7551",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "def train_network(model,optimizer,criterion,X_train,y_train,X_test,y_test,num_epochs,train_losses,test_losses):\n",
    "    for epoch in range(num_epochs):\n",
    "        optimizer.zero_grad()\n",
    "        output_train = model(X_train)\n",
    "        loss_train = criterion(output_train, y_train)\n",
    "        loss_train.backward()\n",
    "        optimizer.step()\n",
    "        output_test = model(X_test)\n",
    "        loss_test = criterion(output_test,y_test)\n",
    "        train_losses[epoch] = loss_train.item()\n",
    "        test_losses[epoch] = loss_test.item()\n",
    "        if (epoch + 1) % 50 == 0:\n",
    "            print(f\"Epoch {epoch+1}/{num_epochs}, Train Loss: {loss_train.item():.4f}, Test Loss: {loss_test.item():.4f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "0140a801-7f7c-4a6e-9612-db1246dba05e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 50/1000, Train Loss: 0.0736, Test Loss: 0.0387\n",
      "Epoch 100/1000, Train Loss: 0.0510, Test Loss: 0.0110\n",
      "Epoch 150/1000, Train Loss: 0.0484, Test Loss: 0.0078\n",
      "Epoch 200/1000, Train Loss: 0.0472, Test Loss: 0.0063\n",
      "Epoch 250/1000, Train Loss: 0.0465, Test Loss: 0.0055\n",
      "Epoch 300/1000, Train Loss: 0.0461, Test Loss: 0.0049\n",
      "Epoch 350/1000, Train Loss: 0.0457, Test Loss: 0.0045\n",
      "Epoch 400/1000, Train Loss: 0.0455, Test Loss: 0.0043\n",
      "Epoch 450/1000, Train Loss: 0.0453, Test Loss: 0.0041\n",
      "Epoch 500/1000, Train Loss: 0.0452, Test Loss: 0.0040\n",
      "Epoch 550/1000, Train Loss: 0.0450, Test Loss: 0.0040\n",
      "Epoch 600/1000, Train Loss: 0.0449, Test Loss: 0.0039\n",
      "Epoch 650/1000, Train Loss: 0.0448, Test Loss: 0.0038\n",
      "Epoch 700/1000, Train Loss: 0.0448, Test Loss: 0.0038\n",
      "Epoch 750/1000, Train Loss: 0.0447, Test Loss: 0.0037\n",
      "Epoch 800/1000, Train Loss: 0.0446, Test Loss: 0.0037\n",
      "Epoch 850/1000, Train Loss: 0.0446, Test Loss: 0.0036\n",
      "Epoch 900/1000, Train Loss: 0.0446, Test Loss: 0.0036\n",
      "Epoch 950/1000, Train Loss: 0.0445, Test Loss: 0.0035\n",
      "Epoch 1000/1000, Train Loss: 0.0445, Test Loss: 0.0035\n"
     ]
    }
   ],
   "source": [
    "iris = load_iris()\n",
    "X = iris['data']\n",
    "y = iris['target']\n",
    "names = iris['target_names']\n",
    "feature_names = iris['feature_names']\n",
    "scaler = StandardScaler()\n",
    "X_scaled = scaler.fit_transform(X)\n",
    "X_train, X_test, y_train, y_test = train_test_split(X_scaled, y, test_size=0.1, random_state=2)\n",
    "X_train = torch.FloatTensor(X_train)\n",
    "X_test = torch.FloatTensor(X_test)\n",
    "y_train = torch.LongTensor(y_train)\n",
    "y_test = torch.LongTensor(y_test)\n",
    "\n",
    "input_dim  = 4 \n",
    "output_dim = 3\n",
    "model = NeuralNetworkClassificationModel(input_dim,output_dim)\n",
    "learning_rate = 0.01\n",
    "criterion = nn.CrossEntropyLoss()\n",
    "optimizer = torch.optim.SGD(model.parameters(), lr=0.1, momentum=0.9)\n",
    "\n",
    "num_epochs = 1000\n",
    "train_losses = np.zeros(num_epochs)\n",
    "test_losses  = np.zeros(num_epochs)\n",
    "train_network(model,optimizer,criterion,X_train,y_train,X_test,y_test,num_epochs,train_losses,test_losses)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "8627007c-3ccc-49a4-9774-96d2a9a759f0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,10))\n",
    "plt.plot(train_losses, label='train loss')\n",
    "plt.plot(test_losses, label='test loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "b9a9e950-abe1-429f-923b-4b4af32c9c51",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training Accuracy: 98.519\n",
      "Test Accuracy: 100.0\n"
     ]
    }
   ],
   "source": [
    "predictions_train = []\n",
    "predictions_test =  []\n",
    "with torch.no_grad():\n",
    "    predictions_train = model(X_train)\n",
    "    predictions_test = model(X_test)\n",
    "def get_accuracy_multiclass(pred_arr,original_arr):\n",
    "    if len(pred_arr)!=len(original_arr):\n",
    "        return False\n",
    "    pred_arr = pred_arr.numpy()\n",
    "    original_arr = original_arr.numpy()\n",
    "    final_pred= []\n",
    "    # we will get something like this in the pred_arr [32.1680,12.9350,-58.4877]\n",
    "    # so will be taking the index of that argument which has the highest value here 32.1680 which corresponds to 0th index\n",
    "    for i in range(len(pred_arr)):\n",
    "        final_pred.append(np.argmax(pred_arr[i]))\n",
    "    final_pred = np.array(final_pred)\n",
    "    count = 0\n",
    "    #here we are doing a simple comparison between the predicted_arr and the original_arr to get the final accuracy\n",
    "    for i in range(len(original_arr)):\n",
    "        if final_pred[i] == original_arr[i]:\n",
    "            count+=1\n",
    "    return count/len(final_pred)\n",
    "\n",
    "train_acc = get_accuracy_multiclass(predictions_train,y_train)\n",
    "test_acc  = get_accuracy_multiclass(predictions_test,y_test)\n",
    "print(f\"Training Accuracy: {round(train_acc*100,3)}\")\n",
    "print(f\"Test Accuracy: {round(test_acc*100,3)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7b9f6874-a87e-4e70-b6c1-fafc9952d5e0",
   "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": "3.9.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}