Search code examples
h2o

Scoring using H2O Flow, on another day


If the goal is to use H2O Flow interface for scoring (which is great BTW), how to accomplish this after a reboot or a system crash?

Is there a way to save models and rerun it, within H2O Flow, without having to train sets again?


Solution

  • The answer is quite simple: - export the model using the UI

    on another day/running a clean instance: - import the model using the UI - import and parse the files used - run predict, select the imported model

    Important: h2o version used on exporting and importing should be the same