yairdata

if my aim is to convert spark ML model to mleap format, do i need to install mleap just on the edge nodes or also on the data nodes ?

unable to create explainer from mleap deserialized pyspark model

4

i have a RandomForestClassificationModel which i deserialized from a saved mleap bundled file: `model = deserialized_pipeline.stages[-1:][0]` gives: `RandomForestClassificationModel (uid=RandomForestClassifier_46858df35b5812fd83ab) with 20 trees` trying to build explainer on the model with:...

great Project ! would be great to upgrade to tensorflow 2 !

the model deal with categorical variables internally. so the wrapper should somehow mimic the conversion of those variables to numeric, isn't it ?