Machine Learning Models on Heart Disease Dataset

This project explores four popular machine learning models: Decision Tree, AdaBoost, Random Forest, and Bagging to predict heart disease using a public dataset (heart.csv). Below is a summary of each model, its implementation, and evaluation results.

1. Decision Tree Classifier

A supervised learning method that splits data into smaller subsets using decision rules. It handles both numerical and categorical data.
Parameters: criterion='gini', max_depth=4, random_state=0
Results: 0.75

2. AdaBoost Classifier

Boosting algorithm that turns weak learners into strong ones by weighting them.
Parameters: n_estimators=100, learning_rate=1.0
Results: AUC Score: 0.81

3.Random Forest Classifier

An ensemble method that builds multiple decision trees and merges their predictions to reduce overfitting.
Parameters: n_estimators=19, max_depth=7, random_state=3
Results: AUC Score: 0.77

4. Bagging Classifier

A technique that trains multiple base learners on random subsets of the training data and aggregates their predictions.
Parameters: n_estimators=20, max_samples=0.8, max_features=10
Results: AUC Score: 0.80

5. Conclusion

The AdaBoost model performed best with an AUC score of 0.81, making it the most effective model for this dataset.