Optimal feature selection with HOEFFDING tree based data classification model for adverse drug reaction

Abstract

Presently, the area of adverse drug reaction (ADR) gains significant interest among researchers due to the rising problems
faced by people. ADR is referred as negative impact occurs in human body by taking drugs irregularly particularly without
physician prescription. Recently, more importance is given to identify the patient populations most at risk, the drugs most
frequently accountable, and the possible reasons of ADRs. The predicting of ADR is challenging due to the presence of
diverse interrelated attributes and the presence of different classes. In this paper, we develop a new Hoeffding tree (HT)
algorithm for the classification of the ADR data. To further improvise the data classification of HT, a particle swarm
optimization (PSO) with simulated annealing (SA) for feature selection process. The presented work is simulated in
MATLAB and the results prove that the presented HT algorithm obtained enhanced results compared to the existing
models under several measures.