Skip to contents

Calculates feature importance for all BAIT model types and groups using standardized coefficients and maximum utility contribution.

feature_importance.Rd

Standardized Coefficients (\(\beta ^{\ast }={\frac {s_{x}}{s_{y}}}\beta\)) as percentage of total.

Usage

feature_importance(
  method = "maximum_utility_contribution",
  modeltype = "binary",
  group = "aggregate"
)

Arguments

method

Method used for calculating feature importance. Either 'maximum_utility_contribution' or 'standardized_coefficients'

modeltype

Model type. Either 'binary' or 'multinomial'

group

Participant group. One of the following 'aggregate', 'aumc', 'olvg', 'intensivists', 'fellows'

Details

Maximum Utility Contribution is calculated by multiplying the criteria level range (difference between the highest and lowest level) with the absolute value of the criterion weight. To calculate a percentage, the maximum utility contributions are summed.

Feature importance is saved for each model as csv files in the inst/extdata folder.

Examples

baitlist::feature_importance(
  method="maximum_utility_contribution",
  modeltype="multinomial",
  group="aggregate"
  )
#> # A tibble: 24 × 6
#>    coefficient_name    coef variable alternative max_util_contrib importance_pct
#>    <chr>              <dbl> <chr>    <chr>                  <dbl>          <dbl>
#>  1 b_expected_los_… -0.0589 expecte… continue              0.0589          0.182
#>  2 b_clinical_situ…  1.25   clinica… continue              1.25            3.87 
#>  3 b_age_continue   -1.85   age      continue              5.56           17.2  
#>  4 b_frailty_conti… -1.08   frailty  continue              2.17            6.72 
#>  5 b_life_expectan…  0.487  life_ex… continue              0.973           3.01 
#>  6 b_suffering_con… -1.14   sufferi… continue              1.14            3.54 
#>  7 b_disability_ca… -0.849  disabil… continue              2.55            7.89 
#>  8 b_disability_pu… -1.33   disabil… continue              2.67            8.26 
#>  9 b_disability_re… -0.379  disabil… continue              0.758           2.35 
#> 10 b_disability_ne… -0.700  disabil… continue              1.40            4.34 
#> 11 b_disability_ga… -0.778  disabil… continue              0.778           2.41 
#> 12 b_family_values… -3.28   family_… continue              3.28           10.1  
#> 13 b_expected_los_… -0.0757 expecte… timelimited           0.0757          0.234
#> 14 b_clinical_situ…  0.282  clinica… timelimited           0.282           0.872
#> 15 b_age_timelimit… -0.971  age      timelimited           2.91            9.02 
#> 16 b_frailty_timel… -0.320  frailty  timelimited           0.641           1.98 
#> 17 b_life_expectan…  0.341  life_ex… timelimited           0.683           2.11 
#> 18 b_suffering_tim… -0.391  sufferi… timelimited           0.391           1.21 
#> 19 b_disability_ca… -0.218  disabil… timelimited           0.655           2.03 
#> 20 b_disability_pu… -0.396  disabil… timelimited           0.792           2.45 
#> 21 b_disability_re… -0.224  disabil… timelimited           0.448           1.39 
#> 22 b_disability_ne… -0.259  disabil… timelimited           0.517           1.60 
#> 23 b_disability_ga… -0.383  disabil… timelimited           0.383           1.19 
#> 24 b_family_values… -1.94   family_… timelimited           1.94            6.00