Mixed credit scoring model of logistic regression and evidence weight in the background of big data

Abstract

Aims of this paper is to amalgamate the logistic regression algorithm under big data with the weight of evidence. To construct a new credit scoring model to analyze the user’s credit score and then divide the user into trustworthy customers and non-trustworthy customers respectively. Calculation of credit score shows the relationship between independent and dependent variable. The weight of the evidence is calculated by the maximum correlation orthogonal transform, which can have a significant effect on models with higher correlation. Due to the error in the data collected, the logistic regression error is large. Therefore, it is suggested that by constructing a hybrid scoring model a more accurate credit score can be obtained. This helps to improve the prediction accuracy of the credit score.