Prediction of Heart Disease Using Random Forest and Rough Set Based Feature Selection

Abstract

Data is generated by the medical industry. Often this data is of very complex nature—electronic records, handwritten scripts, etc.—since it is generated from multiple sources. Due to the Complexity and sheer volume of this data necessitates techniques that can extract insight from this data in a quick and efficient way. These insights not only diagnose the diseases but also predict and can prevent disease. One such use of these techniques is cardiovascular diseases. Heart disease or coronary artery disease (CAD) is one of the major causes of death all over the world. Comprehensive research using single data mining techniques have not resulted in an acceptable accuracy. Further research is being carried out on the effectiveness of hybridizing more than one technique for increasing accuracy in the diagnosis of heart disease. In this article, the authors worked on heart stalog dataset collected from the UCI repository, used the Random Forest algorithm and Feature Selection using rough sets to accurately predict the occurrence of heart disease