MLPredictiveModels 0.4

Educational regression models built from scratch in Python

LinearRegression From Scratch

A minimal, easy-to-understand implementation of Linear Regression in Python using only NumPy.
Designed for educational purposes to show how gradient descent optimizes linear models step by step.

🚀 Features

• Fits simple linear models with gradient descent
• Clean OOP class structure
• Pure NumPy implementation (no scikit-learn)
• Supports multi-feature data
• Easy to extend for regularization
• Printed training loss progress
• Simple fit, predict, and evaluate interface

---

📦 Installation

You can simply copy the linear_regression.py file into your project.
No external dependencies are required except NumPy.

import numpy as np from linear_regression import LinearRegression

toy dataset

X_train = np.array([[1], [2], [3], [4], [5]]) y_train = np.array([2, 4, 6, 8, 10])

initialize model

model = LinearRegression(learning_rate=0.01, epochs=1000)

train

model.fit(X_train, y_train)

predict

predictions = model.predict(X_train)

evaluate

mse, r2 = model.evaluate(X_train, y_train)

print("Predictions:", predictions) print(f"MSE: {mse:.4f}, R2: {r2:.4f}")