Performance Evaluation and Comparative Analysis of Different Machine Learning Algorithms in Predicting Cardiovascular Disease

This study focuses on investigating the performance of different machine learning algorithms and corresponding comparative analysis in predicting cardiovascular disease. Globally this fatal disease causes a plethora of mortality among mankind and so, machine learning algorithms can play a significant role in early detection which will ensure proper treatment for the patients and reduce severity in many cases. The University of California, Irvine (UCI) data repository is utilized for the training and testing of the model. Twelve machine learning algorithms were studied and the performances were observed for default hyperparameter (DHP), grid search cross validation (GSCV) and random search cross validation (RSCV) method. Moreover, computational time were also calculated for both GSCV and RSCV. An accuracy of 92% has been found in both hard and soft voting ensemble classifiers (EVCH and EVCS). However, it observed that Adaboost algorithm outperforms EVCH and EVCS in terms of precision and specificity . Hence, the overall comparative analyses among all the algorithms are carried out extensively where accuracy, precision, sensitivity, specificity, F1 score, and ROC-AUC are brought into action. Jupyter notebook 6.0.3 is utilized for simulation.