Random forest classifier for high entropy alloys phase diagnosis

Abstract

The random forest (RF) algorithm is considered as a powerful statistical classifier that is more popular in other fields but is relatively unknown in HEA(s)’s prediction phase. In this research, Random Forest (RF) technique is used to investigate phase selection principles effectively utilizing a large experimental case study on 401 distinct HEAs, comprising 174 \(SS\), 54 \(IM\), and 173 \(SS+IM\) phases. The accuracy of the proposed method is almost 10% higher than SVM and KNN for classifying HEA(s). Moreover, the precision of the proposed method is similar to ANN. Experimental results indicate the validity and reliability of the RF-based diagnosis method.