A Note on the Comparison of Logit and Discriminant Models of Consumer Credit Behavior
提出用最大似然估计的Logit模型替代线性判别模型进行信用评分，并通过评分实验比较两者表现。
Since the early work of Durand (1941), there has been considerable interest in using quantitative models of consumer credit behavior for credit-granting decisions. Most models are based on the concept of “scoring” by use of weights usually determined as statistically significant coefficients of some linear statistical model, frequently the linear discriminant model. It is the purpose of this note, however, to propose maximum likelihood estimation of the logit model as an alternative, and to compare the two models in a “scoring experiment.”