psi_df: psi_df

In nguyenngocbinh/scorecard.sea: An implementation of score card

Description

Compute the psi for all dataset

Usage

psi_df(dat_train, dat_test, var_skip = NULL)

Arguments

var_skip	Name of force kept variables, Defaults to NULL
df_train	expected or training variable, should be a factor
df_test	observed or actual variable, should be a factor

Details

Check whether the population on which the model is currently being used is similar to the population that was used to develop it.

Value

psi value

Author(s)

Nguyen Ngoc Binh [e-mail]

References

P395, Credit risk analytics - Measurement technique, applications and examples Rules of thumb when using psi:

PSI	shift
-----------------	----------------
< 0.1	no significant shift (green traffic light)
0.1 to 0.25	moderate shift (yellow traffic light)
> 0.25	significant shift (red traffic light)

Examples

data('hmeq')
library(rsample)
hmeq1 <- hmeq[,c("bad", "reason", "job")]
spl <- rsample::initial_split(hmeq1)
train_df <- training(spl)
test_df <- testing(spl)
psi_df1 <- psi_df(train_df, test_df)

nguyenngocbinh/scorecard.sea documentation built on Jan. 4, 2020, 11:43 a.m.