feature_importance.Rd
Compute feature importance
feature_importance( training_data, target_colname, filter_methods = c("FSelectorRcpp_information.gain"), .iters = 500, .first = 30, .split = 0.8, .stratify = TRUE, .seed = 1789 )
training_data | a `data.frame` |
---|---|
target_colname | `character` the name of the column containing the target (output) |
filter_methods | `list` of `character` accepted by `mlr::generateFilterValuesData()` |
.iters | `numeric` number of iterations for the subsampling, default to 500 |
.first | `numeric`, number of feature to display, default to 30 |
.split | `numeric`, ratio of the subsampling splitting ratio, default to 0.8 |
.stratify | `logical`, should the subsampling be stratified, default to `TRUE` |
.seed | `numeric`, fix seed for reproducible example |
a list with two elements a list of `data.frame` for each method and a list of `ggplot` objects