Developing and Evaluating Data-Driven Heart Disease Prediction Models by Ensemble Methods on Different Data Mining Tools

Heart disease stands as a critical global health issue contributing towards the global mortality rate in a significant manner. By extracting patterns and relationships from medical data or clinical records, machine learning techniques can serve as decision support tools for the healthcare providers in predicting heart disease. This study focuses on developing and evaluating prediction models by applying a powerful machine learning technique, ensemble modelling, on different data mining tools, i.e., WEKA and Orange. For analyzing large volumes of historical data to extract meaningful features and insights, these two tools are widely popular. Ada Boost and Gradient Boosting techniques have been applied here for predicting the death rate of heart failure patients. Using a heart failure prediction dataset, the performance metrics have been determined and compared for understanding which model has higher performance on which platform. The experimental result presents classification accuracy, precision, recall and confusion matrix. Such comparative study derived from ensemble modelling on both data mining tools can assist medical professionals and data scientists to understand the performance of the models and differentiate between the tools while choosing for heart disease diagnosis and prediction.