Note: This project was prepared as the final project for the course Data Analytics (Spring 2025) at Constructor University, as part of the Master's in Data Science for Society and Business program.
Comparison on Predictive vs. Generative Models
This project aims to classify whether a patient is likely to have heart disease using both predictive and generative modeling approaches. The study compares traditional machine learning models with a neural network trained in PyTorch, focusing on performance evaluation and model interpretability using partial dependence analysis.
Goal: Predict the presence of heart disease using patient-level features
🔍 The neural network model outperformed other models slightly, especially in precision and ROC-AUC, making it more effective in reducing false positives.