Effect of Different Oversampling Techniques to Handle Class Imbalance Challenges in Coronary Heart Disease Prediction

CHD or Coronary Heart Disease has a considerable influence on global mortality rate, hence ahead of time and precise prediction is crucial for prompt treatment and avoid patient deaths. However, predicting CHD is difficult since the dataset is skewed. To address this challenge, several oversampling techniques are proposed, including Synthetic Minority Oversampling Technique (SMOTE), random oversampling, and Adaptive Synthetic (ADASYN). Incorporating these strategies lowers classifier predisposition towards the majority group while emphasizing proper knowledge of minority notions. Most algorithms perform better on balanced data rather than imbalanced data.This article tests six machine learning models using the CDG dataset. Because this is an imbalanced data problem, accuracy is not used as the primary metric. F1-score is the primary metric. The Extra Trees Classifier achieves an F1-score of 95% by employing random oversampling techniques. F1-score, accuracy, precision, Matthews Correlation Coefficient(MCC) and recall are used for measuring the performance of each algorithm.