Principal Component Analysis Aware BP Neural Network for Personal Information Prediction in Internet Based Services

With the development of Internet based services, the requirement of keeping keep their vitality and the user viscosity has become an important challenge. Better understanding of users behaviour is an effective way to improve the services lifecycle management. As such analysis of users experience from web log, questionnaire and some other ways have been attached much importance. From previous studies it is realised that users personal information is a key data for such analysis. However, due to privacy protection or other security reasons, it is difficult to obtain the users personal profiling. In this research we propose a classification method to predict users age and activity by analysing their questionnaires on certain App services. BP neural network classification approach is employed to this end. We further adopt principal component analysis (PCA) to treat the input data before the predicting model´s training process. The experimental study of proposed method on WeChat payment user experience rating data has shown its possible potential in improving the classifying prediction accuracy.