Machine Learning Models from Scratch

A collection of beginner-friendly machine learning algorithms implemented from scratch using NumPy.

The purpose of this repository is to learn how machine learning algorithms work internally rather than relying on high-level libraries.

Models Included

1. Linear Regression

A regression algorithm used to predict continuous numerical values.

Concepts covered:

• Gradient Descent
• Mean Squared Error (MSE)
• Weight and Bias Optimization

2. Logistic Regression

A classification algorithm for predicting binary outcomes.

Concepts covered:

• Sigmoid Function
• Binary Classification
• Cross-Entropy Loss

3. K-Nearest Neighbors (KNN)

A simple instance-based learning algorithm that classifies data based on the nearest neighbors.

Concepts covered:

• Euclidean Distance
• Majority Voting
• Distance-Based Learning

Technologies Used

• Python
• NumPy

Project Goal

This repository focuses on understanding the mathematics and logic behind machine learning algorithms by implementing them from scratch without using Scikit-Learn.

Future Improvements

• Performance Metrics
• Data Visualization
• Model Comparison
• Additional Algorithms

Author

Adnan Saqib ML Engineer (Learning Journey)