Credit scoring using logistic regression method

Since credit industry has been experiencing intense competitions during these few years, credit scoring, a commonly used quantitative technology that evaluates credit risks of customers by scores, has become increasingly important. Logistic regression is an effective method in credit scoring due to its relative simplicity and accuracy. In this paper, logistic regression is used to build a credit scoring model based on client information data from a commercial bank. As the model reveals, eight variables, including personal salary, present housing status, marriage status, education level, title of a technical position, age, registered permanent residence and industry prove to be statically significant in credit scoring. This conclusion can help the commercial bank make better decisions in two ways: first, it can select good clients by modifying the critical value based on its risk preference, and second, the bank can adjust the weight of some variables in its existing criterion.