RapidMiner

RapidMiner

Discretization

Discretization

I use the freq discretization operator and save a preprocessing model that includes normalization as well with my training set of data. the freq discretization produces 5 bins with ones tailed with infinite values. However whenever I load this preprocess model and apply it to test data ( or even reapply it to the training data itself! ) it puts values in the bins on the end with "?" instead of the infinite ranges, this crashes the learner then with index array problems. I think when loading a model the infinite and -infinite are not be handled correctly
1 REPLY
Moderator

Re: Discretization

Hello,

at the moment at least I have no clue what exactly causes the error. But we will have a look on your problem and keep you informed if we find a solution.

Could you please post your process setup you used to write and apply your preprocessing model?!

Regards,
Tobias