An effective call prediction model based on noisy mobile phone data

Noisy instance in mobile phone data is an important issue for modeling user phone call behavior, with many potential negative consequences. The accuracy of prediction may decrease, thereby increasing the complexity of inferred models and the number of training samples needed. In this paper, we present an effective phone call prediction model based on noisy mobile phone data in order to improve the prediction accuracy for individual mobile phone users. Experimental results on the real phone call log datasets show the effectiveness of our prediction model for individual mobile phone users.