Prediction of Heart Failure Using Naive Bayes Classifiers

A machine learning project that implements a Naive Bayes Classifier from scratch to predict heart failure based on medical data.

🎯 Project Overview

This project demonstrates the mathematical foundations of Naive Bayes classification by building the algorithm entirely from scratch, without relying on pre-built machine learning libraries for the core classification logic.

🔬 What I Built

• Custom Naive Bayes Implementation: Developed the complete Naive Bayes algorithm from mathematical principles
• Dual Data Type Support: Handles both categorical and numerical features simultaneously
• Feature Engineering: Implemented separate probability calculations for different data types
• Performance Metrics: Achieved 88.4% accuracy on heart disease prediction

📊 Feature Overview

🛠️ Technical Implementation

The classifier processes:

• Numerical features: Uses normal distribution assumptions with mean and standard deviation
• Categorical features: Calculates conditional probabilities for each category
• Combined prediction: Merges probabilities from both feature types for final classification

📊 Results

• Accuracy: 88.4%
• Algorithm: Naive Bayes Classifier (implemented from scratch)
• Dataset: Heart disease prediction with mixed categorical and numerical features

📁 Project Structure

• Naive Bayes Classifier.py - Main implementation
• Data/ - Heart disease datasets
• Feature processing.ipynb - Data preprocessing
• Graph/Data representation/ - Visualization notebooks

This project serves as both a practical heart disease prediction tool and an educational resource for understanding the mathematical foundations of Naive Bayes classification.