Author(s)

Zhenming Hu

Corresponding Author

Zhenming Hu

ABSTRACT

This paper adopts the machine learning algorithm to construct a scientific credit scoring method for the credit evaluation of commercial bank customers, and combines the results of the model to analyze the pros and cons of the algorithm. In the empirical analysis of this paper, the data preprocessing is first carried out. Refer to the relevant literature to screen out some variables to participate in modeling. Then the missing value processing and the outlier filling are performed. After that, this paper discusses three algorithms, logistic regression, KNN and GaussianNB, and their built models. The modeling results show that the KNN algorithm takes a long time and is not recommended to be applied under actual conditions. The logistic regression model has applicable conditions. Over-fitting will occur when the conditions are not met. The GaussainNB algorithm performs best.

KEYWORDS

Logistic regression, KNN, GaussainNB, credit evaluation model