loop_calculate_pd_vimp: Convenience function to calculate variable importance over...
In breather/brightbox: Peek into any blackbox learner (including ensembles)

Description Usage Arguments Value Examples

Variable importance is calculated as the difference between the min and max of the partial dependency output for a given feature

1	loop_calculate_pd_vimp(pd_list, vimp_colname = "ensemble")

`pd_list`	output from `loop_calculate_partial_dependency`
`vimp_colname`	name of model (taken from the column names in `pd`) for which to calculate variable importance

a list the length of pd_list, containing the variable importance of each feature

## Not run: 
# Example output from loop_calculate_partial_dependency
pd <- list(a = data.table(feature = c("a", "a", "a"),
                          feature_val = c(1, 3.5, 6),
                          model1 = c(-2.5, 0, 2.5),
                          model2 = c(0, 0, 0),
                          ensemble = c(-2.5, -0.75, 0)),
           b = data.table(feature = c("b", "b", "b"),
                          feature_val = c(2, 3, 4),
                          model1 = c(0, 0, 0),
                          model2 = c(0, 0, 0),
                          ensemble = c(1, 2, 3)))
loop_calculate_pd_vimp(pd, vimp_colname = "ensemble")

## End(Not run)