Merge method for EBM objects — merge.EBM • ebm

Merge multiple EBMs together.

Usage

# S3 method for EBM
merge(x, y, ...)

Arguments

x, y: Fitted ebm objects that have been trained on similar data sets that have the same set of features.
...: Additional ebm objects to be merged.

Value

A merged ebm object.

Note

As of right now, the merge() function produces the following error message:

Error in py_repr(x) :
  AttributeError: 'ExplainableBoostingRegressor' object has no attribute 'cat_smooth'
Run `reticulate::py_last_error()` for details.

This seems to be a bug in the underlying interpret library and does not prevent this function from working. The error message is seemingly just a side effect.

Examples

if (FALSE) {
# Generate list of EBMs with different random seeds
ebms <- lapply(1:3, FUN = function(i) {
  ebm(mpg ~ ., data = mtcars, outer_bags = 1, random_state = i, obj = "rmse")
})

# Merge EBMs into one and plot term contribution for `cyl`
merged <- do.call(merge, args = ebms)
plot(merged, term = "cyl")
}