[Question] What is the best way to use MLeap XGBoost in python?

combust / mleap

MLeap: Deploy ML Pipelines to Production

https://combust.github.io/mleap-docs/

Apache License 2.0

1.5k stars 310 forks source link

[Question] What is the best way to use MLeap XGBoost in python? #557

Open tinaxi opened 5 years ago

tinaxi commented 5 years ago

I have a model trainer python project that needs support for serializing XGBoost models. I am trying to call the Scala APIs via the jvm following the XGBoostClassificationModelParitySpec example, but there are a bunch of implicit classes that cannot be found this way in python and I cannot import the ml.combust.mleap.spark.SparkSupport._ via jvm. Do you have any suggestions about how to integrate MLeap XGBoost in python? Any help would be really appreciated!

ancasarb commented 5 years ago

At the moment, we have support for xgboost in Scala, haven't tried it from python. There's some discussion on supporting custom transformers from python, which I'm not sure, but may help https://github.com/combust/mleap/issues/552.