error in ceteris_paribus.default with a response with several labels

[Quentin62]

When you have a response variable with several labels and provide a data.frame with several rows in ceteris_paribus, it returns an error like this:

Error in `$<-.data.frame`(`*tmp*`, "_ids_", value = c(1L, 1L, 1L, 1L,  : 
  replacement has 400 rows, data has 80

This can be corrected with the following change:
In the file ceteris.paribus.R, line 146 is:
new_observation_ext$`_ids_` <- rep(1:nrow(new_observation), each = nrow(new_observation))

it should be:
new_observation_ext$`_ids_` <- rep(1:nrow(new_observation), each = length(col_yhat))

Toy example:

library("randomForest")
library("DALEX")

titanic <- na.omit(titanic)
# we predict embarked instead of survived for generating the bug
model_titanic_rf <- randomForest(embarked ~ gender + age + class + survived +
                                   fare + sibsp + parch,  data = titanic)
model_titanic_rf

explain_titanic_rf <- explain(model_titanic_rf,
                              data = titanic[,-4],
                              y = titanic$survived,
                              label = "Random Forest v7")

# select few passangers
selected_passangers <- select_sample(titanic, n = 20)
cp_rf <- ceteris_paribus(explain_titanic_rf, selected_passangers)
cp_rf

[pbiecek]

Thanks, fixed