calculate_weight: Calculated empirical density and weight based on variable...

In ModelOriented/vivo: Variable Importance via Oscillations

Description

This function calculate an empirical density of raw data based on variable split from Ceteris Paribus profiles. Then calculated weight for values generated by DALEX::predict_profile(), DALEX::individual_profile() or ingredients::ceteris_paribus().

Usage

calculate_weight(profiles, data, variable_split)

Arguments

profiles	data.frame generated by DALEX::predict_profile(), DALEX::individual_profile() or ingredients::ceteris_paribus()
data	data.frame with raw data to model
variable_split	list generated by vivo::calculate_variable_split()

Value

Return an weight based on empirical density.

Examples

library("DALEX", warn.conflicts = FALSE, quietly = TRUE)
data(apartments)

split <- vivo::calculate_variable_split(apartments,
                        variables = colnames(apartments),
                        grid_points = 101)

library("randomForest", warn.conflicts = FALSE, quietly = TRUE)
apartments_rf_model <- randomForest(m2.price ~ construction.year + surface +
                                    floor + no.rooms, data = apartments)

explainer_rf <- explain(apartments_rf_model, data = apartmentsTest[,2:5],
                        y = apartmentsTest$m2.price)

new_apartment <- data.frame(construction.year = 1998, surface = 88, floor = 2L, no.rooms = 3)

profiles <- predict_profile(explainer_rf, new_apartment)

library("vivo")
calculate_weight(profiles, data = apartments[, 2:5], variable_split = split)

ModelOriented/vivo documentation built on Sept. 29, 2020, 10:53 p.m.