does RegressionTree.fit see the predictors as nominal variables or ordinal variables if you set 'CategoricalPredictors' to 'all'?

As the title suggests I am curious to find out how MATLAB "views" the predictors if you state them as categorical? I have nominal predictors but I don't know if MATLAB sees them as such? Or maybe it doesn't matter as far as the resulting tree is concerned?

Thank you!

Answer 1

[Ilya]

For trees, "categorical predictor" means "nominal predictor". If you had an ordinal predictor, you could just use it as is because distance between ordinal levels does not affect the tree construction.