Credit risk classifications of e-commerce based on KPCA-MPSO-ANNAuthor(s): JianpingWu , Bangzhu Zhu
The paper attempts to classify E-commerce credit risk in a clearer way by adopting KPCA, MPSO and ANN. In the KPC classification, the data was pre-processed in the first place, and then the eigenvalues and eigenvectors were extracted to reduce the dimensions of the E-commerce credit risk. Furthermore, the study searched the inertia weight and threshold of BP neutral network through the improved MPSO, and determined the inertia weight and threshold value BP neural network. The data of 13 enterprises was trained first, and that of another 5 enterprises was tested and predicted. And finally, the result was classified. The study proved that the KPCA-MPSO-ANN based analysis was quite effective, providing a sound basis, reference and empirical case for classification and evaluation of E-commerce enterprises. Besides, it is of some help to promote the development of E-commerce industry.