Get predictive indicator for partial models given a model

Usage

model_partials(model, newdata = NULL, verbose = TRUE)

Arguments

model: model
newdata: Optional data frame
verbose: verbose

Examples


data("credit_woe")

m <- glm(bad ~ age_woe + flag_res_phone_woe + months_in_the_job_woe +
   payment_day_woe + sex_woe + profession_code_woe + marital_status_woe,
   family = binomial, data = head(credit_woe, 10000)
   )

model_partials(m)
#> ℹ Fitting and evaluating model with 1 variables: age_woe
#> ℹ Creating woe binning ...
#> ℹ Fitting and evaluating model with 2 variables: age_woe, flag_res_phone_woe
#> ℹ Creating woe binning ...
#> ℹ Fitting and evaluating model with 3 variables: age_woe, flag_res_phone_woe, m...
#> ℹ Creating woe binning ...
#> ℹ Fitting and evaluating model with 4 variables: age_woe, flag_res_phone_woe, m...
#> ℹ Creating woe binning ...
#> ℹ Fitting and evaluating model with 5 variables: age_woe, flag_res_phone_woe, m...
#> ℹ Creating woe binning ...
#> ℹ Fitting and evaluating model with 6 variables: age_woe, flag_res_phone_woe, m...
#> ℹ Creating woe binning ...
#> ℹ Fitting and evaluating model with 7 variables: age_woe, flag_res_phone_woe, m...
#> ℹ Creating woe binning ...
#> # A tibble: 7 × 5
#>   variable                 ks   auc    iv  gini
#>   <fct>                 <dbl> <dbl> <dbl> <dbl>
#> 1 age_woe               0.204 0.624 0.190 0.248
#> 2 flag_res_phone_woe    0.207 0.641 0.248 0.283
#> 3 months_in_the_job_woe 0.223 0.650 0.276 0.301
#> 4 payment_day_woe       0.238 0.656 0.304 0.312
#> 5 sex_woe               0.247 0.659 0.309 0.319
#> 6 profession_code_woe   0.244 0.661 0.329 0.321
#> 7 marital_status_woe    0.244 0.662 0.335 0.324

model_partials(m, newdata = tail(credit_woe, 10000))
#> ℹ Fitting and evaluating model with 1 variables: age_woe
#> ℹ Creating woe binning ...
#> ℹ Creating woe binning ...
#> ℹ Fitting and evaluating model with 2 variables: age_woe, flag_res_phone_woe
#> ℹ Creating woe binning ...
#> ℹ Creating woe binning ...
#> ℹ Fitting and evaluating model with 3 variables: age_woe, flag_res_phone_woe, m...
#> ℹ Creating woe binning ...
#> ℹ Creating woe binning ...
#> ℹ Fitting and evaluating model with 4 variables: age_woe, flag_res_phone_woe, m...
#> ℹ Creating woe binning ...
#> ℹ Creating woe binning ...
#> ℹ Fitting and evaluating model with 5 variables: age_woe, flag_res_phone_woe, m...
#> ℹ Creating woe binning ...
#> ℹ Creating woe binning ...
#> ℹ Fitting and evaluating model with 6 variables: age_woe, flag_res_phone_woe, m...
#> ℹ Creating woe binning ...
#> ℹ Creating woe binning ...
#> ℹ Fitting and evaluating model with 7 variables: age_woe, flag_res_phone_woe, m...
#> ℹ Creating woe binning ...
#> ℹ Creating woe binning ...
#> # A tibble: 14 × 6
#>    variable              sample    ks   auc    iv  gini
#>    <fct>                 <fct>  <dbl> <dbl> <dbl> <dbl>
#>  1 age_woe               train  0.204 0.624 0.190 0.248
#>  2 age_woe               test   0.194 0.624 0.213 0.249
#>  3 flag_res_phone_woe    train  0.207 0.641 0.248 0.283
#>  4 flag_res_phone_woe    test   0.208 0.645 0.262 0.289
#>  5 months_in_the_job_woe train  0.223 0.650 0.276 0.301
#>  6 months_in_the_job_woe test   0.225 0.654 0.312 0.307
#>  7 payment_day_woe       train  0.238 0.656 0.304 0.312
#>  8 payment_day_woe       test   0.234 0.658 0.335 0.317
#>  9 sex_woe               train  0.247 0.659 0.309 0.319
#> 10 sex_woe               test   0.241 0.665 0.344 0.329
#> 11 profession_code_woe   train  0.244 0.661 0.329 0.321
#> 12 profession_code_woe   test   0.249 0.669 0.365 0.337
#> 13 marital_status_woe    train  0.244 0.662 0.335 0.324
#> 14 marital_status_woe    test   0.251 0.671 0.385 0.341