一种基于贝叶斯判别的信用评分方法

A Credit Scoring Technology Based on Bayes Discriminant Analysis

摘要: 本文借鉴了FICO评分的思想，基于贝叶斯判别定理推导出一套评分模型，评分模型最终为一个目标函数是线性函数，约束条件含有二次等式约束的最优化问题。最后，通过一个实例与Logistic回归做了对比，实例结果表明模型是有效的，且模型能够更好的支持实际业务应用场景。

Abstract: According to FICO Score theory and Bayes Discrimination, the credit scoring model is derived, which ends up as an optimization model with linear objective function and quadratic equality constraints. Finally, compared with Logistic Regression through an example, the result shows that the credit scoring model is effective and can support application scenarios of practical business better.