In bcjaeger/aorsf: Accelerated Oblique Random Forests

library(magrittr)

You can compute individual conditional expectation and individual conditional expectations in three ways:

• using in-bag predictions for the training data. In-bag individual conditional expectation indicates relationships that the model has learned during training. This is helpful if your goal is to interpret the model.

• using out-of-bag predictions for the training data. Out-of-bag individual conditional expectation indicates relationships that the model has learned during training but using the out-of-bag data simulates application of the model to new data. This is helpful if you want to test your model's reliability or fairness in new data but you don't have access to a large testing set.

• using predictions for a new set of data. New data individual conditional expectation shows how the model predicts outcomes for observations it has not seen. This is helpful if you want to test your model's reliability or fairness.

Classification

Begin by fitting an oblique classification random forest:

Compute individual conditional expectation using out-of-bag data for flipper_length_mm = c(190, 210).

pred_spec <- list(flipper_length_mm = c(190, 210))

ice_oob <- orsf_ice_oob(fit_clsf, pred_spec = pred_spec)

ice_oob

There are two identifiers in the output:

• id_variable is an identifier for the current value of the variable(s) that are in the data. It is redundant if you only have one variable, but helpful if there are multiple variables.

• id_row is an identifier for the observation in the original data.

Note that predicted probabilities are returned for each class and each observation in the data. Predicted probabilities for a given observation and given variable value sum to 1. For example,

ice_oob %>%
 .[flipper_length_mm == 190] %>% 
 .[id_row == 1] %>% 
 .[['pred']] %>% 
 sum()

Regression

Begin by fitting an oblique regression random forest:

Compute individual conditional expectation using new data for flipper_length_mm = c(190, 210).

pred_spec <- list(flipper_length_mm = c(190, 210))

ice_new <- orsf_ice_new(fit_regr, 
                        pred_spec = pred_spec,
                        new_data = penguins_orsf_test)

ice_new

You can also let pred_spec_auto pick reasonable values like so:

pred_spec = pred_spec_auto(species, island, body_mass_g)

ice_new <- orsf_ice_new(fit_regr, 
                        pred_spec = pred_spec,
                        new_data = penguins_orsf_test)

ice_new

By default, all combinations of all variables are used. However, you can also look at the variables one by one, separately, like so:

ice_new <- orsf_ice_new(fit_regr, 
                        expand_grid = FALSE,
                        pred_spec = pred_spec,
                        new_data = penguins_orsf_test)

ice_new

And you can also bypass all the bells and whistles by using your own data.frame for a pred_spec. (Just make sure you request values that exist in the training data.)

custom_pred_spec <- data.frame(species = 'Adelie', 
                               island = 'Biscoe')

ice_new <- orsf_ice_new(fit_regr, 
                        pred_spec = custom_pred_spec,
                        new_data = penguins_orsf_test)

ice_new

Survival

Begin by fitting an oblique survival random forest:

Compute individual conditional expectation using in-bag data for bili = c(1,2,3,4,5):

ice_train <- orsf_ice_inb(fit_surv, pred_spec = list(bili = 1:5))
ice_train

If you don't have specific values of a variable in mind, let pred_spec_auto pick for you:

ice_train <- orsf_ice_inb(fit_surv, pred_spec_auto(bili))
ice_train

Specify pred_horizon to get individual conditional expectation at each value:

ice_train <- orsf_ice_inb(fit_surv, pred_spec_auto(bili),
                          pred_horizon = seq(500, 3000, by = 500))
ice_train

Multi-prediction horizon ice comes with minimal extra computational cost. Use a fine grid of time values and assess whether predictors have time-varying effects.

bcjaeger/aorsf documentation built on April 3, 2025, 4:16 p.m.