Comparative Analysis of Classification Algorithms for Cardiovascular Disease Dataset with WEKA

Abstract

Predicting Cardiovascular Disease is crucial in healthcare, with machine learning algorithms playing a key role in identifying high-risk patients. Cardiovascular diseases are the leading cause of death globally, highlighting the urgent need for innovative diagnostic tools to predict cardiovascular disease more accurately and efficiently. Through rigorous experimentation across diverse data partitions, we seek optimal hyper-parameter configurations to enhance model accuracy and robustness. This study evaluates the performance of four machine learning classification algorithms namely Multilayer Perceptron, Random Forest, J48, Sequential Minimal Optimization on a cardiovascular disease dataset. Percentage splits from 60% to 95% in 5% increments were employed, with 95% yielding the highest observed accuracy under this experimental setting. Our findings highlight the importance of optimal parameter selection and provide valuable insights into each algorithm, enhancing predictive accuracy in cardiovascular diagnosis.