Prior studies show that LR has better prediction power than traditional approaches. NN is a useful statistical technique that has massive computing power, powerful memory, learning ability, and fault-tolerance ability. DA takes into account multiple predictors and determines which predictor has discriminative power in a more objective sense. However, the values of independent variables in the model must have a normal distribution. As the fi nancial data of banks used in this chapter do not conform to a normal distribution, we choose to employ LR that better fi ts the non-normal variables.

38.3.1 Logistic Regression