Integer vectors are quite problematic because sometimes they might be treated as a numeric variable (to they should be discretized before calculating the information gain), however sometimes they might be used to encode a factor variable (e.g., days of a week, gender, and so on), so discretizing them is not a good idea;)
I'll add integer2numeric to control this behavior, and inform the users that it might be a problem because it seems that they're not aware of that.
Also, it makes the package a bit more consistent because right now information_gain leaves integer columns as is, but discretize discretizes them:(
Integer vectors are quite problematic because sometimes they might be treated as a numeric variable (to they should be discretized before calculating the information gain), however sometimes they might be used to encode a factor variable (e.g., days of a week, gender, and so on), so discretizing them is not a good idea;)
I'll add
integer2numeric
to control this behavior, and inform the users that it might be a problem because it seems that they're not aware of that.Also, it makes the package a bit more consistent because right now
information_gain
leaves integer columns as is, butdiscretize
discretizes them:(