Adjusted Empirical Estimate of Information Value for Credit Scoring Models

Warning

This publication doesn't include Faculty of Arts. It includes Faculty of Science. Official publication website can be found on muni.cz.
Authors

ŘEZÁČ Martin KOLÁČEK Jan

Year of publication 2011
Type Article in Proceedings
Conference PROCEEDINGS ASMDA 2011
MU Faculty or unit

Faculty of Science

Citation
Field Applied statistics, operation research
Keywords Credit scoring; Quality indexes; Information value; ESIS
Description To measure the quality of scoring models it is possible to use quantitative indexes such as Gini index, K-S and Information value. The paper deals with the Information value, mainly with issues connected to its computation. The classical way of computation, i.e. empirical estimate using deciles of scores, is easy to implement, but may lead to strongly biased results. Kernel estimate or empirical estimates with supervised interval selection (ESIS) seems to be more appropriate to use. The main contribution of this paper is a proposal of an adjusted procedure for estimation of the Information value. It is based on ideas of ESIS with adjustment for choice of required number of observations in constructed intervals. The properties of all listed Information value estimators are discussed in the simulation study.
Related projects:

You are running an old browser version. We recommend updating your browser to its latest version.