Random Forest Classifier based on Variable Precision Rough Set Theory

Abstract

Decision-making process is supported by Machine learningbased classification techniques in many areas of health care. Classification performance of decision system can be improved using the attribute reduction mainly in the situation of high data dimensionality dilemma .This paper proposes, Random forest Classifier (RFC) approach which is based on the Variable Precision Rough Set (VPRS) theory. The first phase of proposed approach focus at attribute reduction of available dataset using VPRS .Directing from dimensionality reduction to predictive model construction, and in next phase, the obtained abridged dataset is provided as the input of RFC to build a more accurate classification model. The performance is evaluated in terms of classification accuracy and time complexity. The experimental results show that the enhanced RFC has higher accuracy and correctly classified instances as compared with the existing algorithms.