Prediction of Heart Disease Using Advanced Stacking and a Comparative Analysis of Feature Selection Techniques

Heart disease remains one of the most significant global health challenges, and the need for accurate and reliable diagnostic tools is constantly increasing. This paper investigates the impact of various feature selection techniques, including PCA, ReliefF, Mutual Information, and Chi-Square, on machine learning models for heart disease prediction. It introduces an Advanced Stacking approach designed to reduce overfitting and enhance accuracy. This multi-layered method leverages out-of-fold predictions to train the base-layers. Using a preprocessed heart disease dataset, a comprehensive comparison is performed across multiple base classifiers and the proposed stacking model. The top result, combining ReliefF with Advanced Stacking, achieved ۸۷% accuracy. The model consistently outperforms individual classifiers, providing a robust and accurate predictive tool.

کلیدواژه ها:

Heart Disease Prediction ، Advanced Stacking ، Feature Selection ، Ensemble Learning ، Predictive Modeling