I have model objects either pickled object or H2O POJO. Is it possible to call those objects and do the scoring in MATLAB?

  • $\begingroup$ Scoring as in - you run the model normally in python, perform the predictions in that code in python and then perform the prediction scoring in MATLAB. (i.e. not the model training with cross-validation in MATLAB). Yup? $\endgroup$
    – grochmal
    Jun 7 '19 at 18:18
  • $\begingroup$ @grochmal Yes indeed I do modelling in python and save my model object there. Then I wanna use this model object in MATLAB to do scoring. $\endgroup$ Jun 8 '19 at 16:14
  • $\begingroup$ Wait, by "scoring" you mean executing the model on new data (that is available only in MATLAB at that point in time), yes? i.e. perform the prediction on data that the python code did not see, or is not able to see. Not "scoring" the model based on predictions it already did. (P.S. "scoring" is quite an overloaded word). $\endgroup$
    – grochmal
    Jun 8 '19 at 23:13
  • $\begingroup$ Same data will be in MATLAB in production and the model object will be used to predict. But this is not relevant to my question. I just wanna learn if I can use my pickled or POJO object, created in pyhton, in MATLAB. $\endgroup$ Jun 11 '19 at 7:51
  • $\begingroup$ you can't by default run one language in another but that does not mean there aren't workarounds. Yet, workarounds may be slow, or moreover, need a good deal of work to not be slow. I'm trying to weight the possibilities (and don't want to give a correct but useless in practice answer). So probably the last one: How fast the resulting classification need to be? $\endgroup$
    – grochmal
    Jun 11 '19 at 11:18
"Data.p" #yours pickle file

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.