Principal Component Analysis And General Regression Auto Associative Neural Network Hybrid As One-Class Classifier

In this paper we develop the principal component analysis (PCA) and general regression auto associative neural network (GRAANN) based hybrid as a one-class classifier (PCA-GRAANN). We test the effectiveness of PCA-GRAANN on bankruptcy prediction datasets namely Spanish banks, Turkish banks, US banks and UK banks; UK credit dataset and the benchmark WBC dataset. When compared the results of another recently proposed hybrid, particle swarm optimization trained auto associative neural network (PSOAANN) [1], PCA-GRAANN yielded mixed results. We conclude that PCA-GRAANN can be used as a viable alternative for any one-class classifier.