Machine Learning from Scratch (Java)

A custom implementation of core Machine Learning algorithms built entirely in Java without external ML libraries.

🎯 Goal

To deconstruct the "black box" of libraries like Scikit-learn and understand the mathematical foundations of algorithms by engineering them from the ground up.

⚡ Algorithms Implemented

• Linear Regression: Using Gradient Descent optimization.
• K-Nearest Neighbors (KNN): Custom distance metric calculations.
• Logistic Regression: Sigmoid activation and cost function minimization.
• (Add whatever else we worked on: Decision Trees, K-Means, etc.)

🛠️ Tech Stack

• Language: Java (OOP principles)
• Math: Linear Algebra & Calculus logic implemented raw.
• Tools: Maven/Gradle (if you used them), JUnit for testing.

---

🧠 What I Learned

Building this from scratch taught me the "magic" behind libraries like Scikit-Learn:

• Math to Code: How to translate mathematical formulas (like Gradient Descent and Euclidean Distance) into efficient Java code.
• OOP Design: structuring a clean architecture where different models (KNN, Linear Regression) share a common interface.
• Data Handling: Writing manual logic to normalize and split data without relying on Pandas.

🚀 Future Improvements

Ways I plan to make this engine better:

• Add More Algorithms: Implement Logistic Regression and a simple Neural Network.
• CSV Support: Build a parser to read real datasets from .csv files automatically.
• Optimization: Improve matrix multiplication performance for larger datasets.

💻 How to Run

This project is built with standard Java (no external ML dependencies).

Option 1: Using Eclipse (Recommended)

1. Clone this repo: git clone https://github.com/YOUR-USERNAME/YOUR-REPO.git
2. Open Eclipse and go to File > Open Projects from File System.
3. Select the project folder.
4. Navigate to src/ml/app/Main.java.
5. Right-click and select Run As > Java Application.

Option 2: Terminal

cd src
javac ml/app/Main.java
java ml.app.Main