Strange results for titanic data

[pbiecek]

Any idea why results for the titanic are so strange?

library("DALEX")
library("randomForest")
load("models/titanic_rf_v6.rda")
load("models/titanic.rda")
explain_titanic_rf <- explain(model = titanic_rf_v6, 
                              data = titanic[,c(1:4,6:8)],
                              y = titanic$survived == "yes", 
                              label = "Random Forest v6")
johny_d <- data.frame(
  class = factor("1st", levels = c("1st", "2nd", "3rd", "deck crew", "engineering crew", 
                                  "restaurant staff", "victualling crew")),
  gender = factor("male", levels = c("female", "male")),
  age = 8,
  sibsp = 0,
  parch = 0,
  fare = 72,
  embarked = factor("Southampton", levels = c("Belfast", "Cherbourg", "Queenstown", "Southampton"))
)

shapper_johny <- shapper::individual_variable_effect(titanic_rf_v6, titanic[,c(1:4,6:8)], DALEX:::yhat.randomForest, johny_d)
plot(shapper_johny)

[maksymiuks]

You have made johny_d with wrong column order, therefore shap returned NA.

We have added exception for such a situation, thanks for that!

[pbiecek]

Thanks, indeed it was the problem. But maybe instead of rising an error you can fix the order of variables, in the same way as it is done for iBreakDown lines 109-112 https://github.com/ModelOriented/iBreakDown/blob/master/R/local_attributions.R#L109 ?

[maksymiuks]

Rising an error fixses a problem when user have made typo.