RegressionLibrary 0.3

Educational regression models built from scratch in Python

RegressionLib

RegressionLib is a lightweight, educational regression library implemented from scratch in Python with NumPy. It supports multiple regression models, aiming to help students and researchers understand the mathematical foundations of regression algorithms without the black-box nature of larger frameworks.

---

🚀 Features

• Linear Regression
• Lasso Regression (L1 regularization)
• Ridge Regression (L2 regularization)
• Polynomial Regression
• Clean, beginner-friendly API (fit, predict, evaluate)
• Minimal dependencies (NumPy only)
• Clear educational codebase, easy to extend import numpy as np from regressionlib import LinearRegression, LassoRegression, RidgeRegression, PolynomialRegression

Example training data

X_train = np.array([[1], [2], [3], [4], [5]]) y_train = np.array([1.5, 2.5, 3.5, 4.5, 5.5])

Linear Regression

lr_model = LinearRegression() lr_model.fit(X_train.flatten(), y_train, learning_rate=0.01, epochs=1000) mse, r2, acc = lr_model.evaluate(X_train.flatten(), y_train) print(f"Linear Regression - MSE: {mse:.4f}, R2: {r2:.4f}, Accuracy: {acc:.2f}%")

Lasso Regression

lasso = LassoRegression(learning_rate=0.01, iterations=1000, l1_penalty=0.1) lasso.fit(X_train, y_train) mse, r2, acc = lasso.evaluate(X_train, y_train) print(f"Lasso Regression - MSE: {mse:.4f}, R2: {r2:.4f}, Accuracy: {acc:.2f}%")

Ridge Regression

ridge = RidgeRegression(learning_rate=0.01, epochs=1000, l2_penalty=0.1) ridge.fit(X_train, y_train) mse, r2, acc = ridge.evaluate(X_train, y_train) print(f"Ridge Regression - MSE: {mse:.4f}, R2: {r2:.4f}, Accuracy: {acc:.2f}%")

Polynomial Regression

poly = PolynomialRegression(degree=2, learning_rate=0.01, epochs=1000) poly.fit(X_train, y_train) mse, r2, acc = poly.evaluate(X_train, y_train) print(f"Polynomial Regression - MSE: {mse:.4f}, R2: {r2:.4f}, Accuracy: {acc:.2f}%")