Weight of Evidence Based Segmentation of a Variable
Segment a numeric variable based on a dichotomous dependent variable by using the weight of evidence (WOE) approach (Ref: Siddiqi, N. (2006) <10.1002>).
The underlying algorithm adopts a recursive approach to create segments that are diverse in respect of their WOE values and meet the demands of user-defined parameters. The algorithm also aims to maintain a monotonic trend in WOE values of consecutive segments. As such, it can be particularly helpful in improving robustness of linear and logistic regression models.10.1002>