A Machine Learning Approach to Heart Disease Prediction: 5-Fold Cross Validation and Hyperparameter Optimization
Authors:-Dr.N.Chandrasekhar
Abstract-The primary objective of this research is to develop an effective predictive model for heart disease using various Machine Learning (ML) algorithms. In this study, four different ML models—Gradient Boosting (GB), Random Forest (RF), LightGBM (LGBM), and AdaBoost (AB)—were implemented and evaluated for their prediction accuracy. To ensure the reliability and generalization of the models, 5-fold cross-validation was applied along with Grid Search Cross Validation (Grid Search CV) for hyperparameter tuning. This technique helped in identifying the optimal parameters for each algorithm, thereby improving their performance. Among all the models, Gradient Boosting achieved the highest accuracy of 95.08%, followed by Random Forest and LightGBM, both with 91.80%, and AdaBoost with 90.16%. These results highlight the effectiveness of ensemble-based ML models, particularly Gradient Boosting, in accurately predicting the risk of heart disease.
