Heart Disease Prediction Using Hybrid Random Forest Model Integrated with Linear Model

Abstract

The objective of the paper is to throw light on few existing heart disease predicting approaches and proposes a Hybrid Random Forest Model Integrated with Linear Model (HRFMILM) for predicting and identifying the HDs at an early stage. Even though the linear model has simple estimation procedure, it is very sensitive to outliers and may lead to overfitting process. On the other hand, averaging in Random Forest Model (RFM) improves the overall accuracy and reduces the possibility of overfitting. The dataset is collected from standard UCI repository. Experimental results concluded that the integration of Linear Model with RFM makes the simple estimation procedure with improved overall accuracy than the respective models. Further, the proposed method compares the prediction performance of few existing approaches in terms of parameters, namely, precision, recall and F1-score.