Effects of Feature Engineering on Machine Learning Model Performances

Abstract

This paper will investigate the effects of feature engineering on the accuracy of the Random Forest Regressor (RFR), Decision Tree Classifier (DTC), and Linear Regressor (LR) models when predicting the presence of a heart attack. By utilizing a tabular dataset of eight heart disease factors, we evaluate the models' accuracy when predicting a binary output relating to the presence of a heart attack. The findings highlight the remarkable potency of the DTC when predicting a binary value using tabular data points. They also highlight the detrimental effects on model accuracy of the incorrect utilization of feature engineering combinations. The valuable insights brought by feature engineering will contribute to the development of informed heart attack prevention measures because high-risk individuals can make informed decisions regarding their lifestyle with the help of accurate models.