Research on the Prediction of Customers’ Arrears Under Big Data Background

Abstract

This paper proposed a new prediction method to predict customers’ arrears by means of multivariable linear regression model and convolutional neural network (CNN) on Hadoop big data platform with mass of customers’ power consumption. The proposed method made the association between different type of customers’ data and related traits with customers’ arrear, and estimated the probability of customers’ arrear based on evaluated customers’ credit level and prediction of customers’ arrear. The application results illustrated that the method is available.