Kolmogorov-Smirnov (KS) statistics is used to assess predictive power for marketing or credit risk models. It is the maximum difference between cumulative event and non-event distribution across score/probability bands. The gains table typically has across score bands and can be used to find the KS for a model.

blr_ks_chart(gains_table, title = "KS Chart", yaxis_title = " ",
  xaxis_title = "Cumulative Population %", ks_line_color = "black")

Arguments

gains_table

An object of class blr_gains_table.

title

Plot title.

yaxis_title

Y axis title.

xaxis_title

X axis title.

ks_line_color

Color of the line indicating maximum KS statistic.

References

https://doi.org/10.1198/tast.2009.08210

https://www.ncbi.nlm.nih.gov/pubmed/843576

See also

Examples

model <- glm(honcomp ~ female + read + science, data = hsb2, family = binomial(link = 'logit')) gt <- blr_gains_table(model) blr_ks_chart(gt)