Linear Regression

1/**2 * Example 07: Linear Regression3 *4 * Build a simple linear regression model to predict continuous values.5 * Learn the basics of supervised learning with Deepbox.6 */78import { mae, mse, r2Score } from "deepbox/metrics";9import { LinearRegression } from "deepbox/ml";10import { tensor } from "deepbox/ndarray";11import { trainTestSplit } from "deepbox/preprocess";1213console.log("=== Linear Regression ===\n");1415// Generate synthetic data for demonstration16// True relationship: y = 2x + 3 + noise17const X_data: number[][] = [];18const y_data: number[] = [];1920for (let i = 0; i < 100; i++) {21  const x = i / 10;22  // Add random noise to make it realistic23  const y = 2 * x + 3 + (Math.random() - 0.5) * 2;24  X_data.push([x]);25  y_data.push(y);26}2728// Convert to tensors29const X = tensor(X_data);30const y = tensor(y_data);3132console.log(`Dataset: ${X.shape[0]} samples, ${X.shape[1]} features\n`);3334// Split data: 80% training, 20% testing35const [X_train, X_test, y_train, y_test] = trainTestSplit(X, y, {36  testSize: 0.2,37  randomState: 42,38});3940console.log(`Training set: ${X_train.shape[0]} samples`);41console.log(`Test set: ${X_test.shape[0]} samples\n`);4243// Create and train the linear regression model44const model = new LinearRegression();45model.fit(X_train, y_train);4647console.log("Model trained!");48console.log(`Coefficients: ${model.coef?.toString()}`);49console.log(`Intercept: ${model.intercept}\n`);5051// Generate predictions on test set52const y_pred = model.predict(X_test);5354// Calculate performance metrics55const r2 = r2Score(y_test, y_pred);56const mseValue = mse(y_test, y_pred);57const maeValue = mae(y_test, y_pred);5859console.log("Model Performance:");60console.log(`R² Score: ${r2.toFixed(4)}`);61console.log(`MSE: ${mseValue.toFixed(4)}`);62console.log(`MAE: ${maeValue.toFixed(4)}`);6364console.log("\n✓ Linear regression complete!");65