Move interpretable logic to InterpretableTabularPredictor #2981
Conversation
|
Job PR-2981-2955ec4 is done. |
Innixma
force-pushed
the
cleanup_interpret
branch
from
2955ec4
to
e607559
Compare
|
Job PR-2981-e607559 is done. |
| @@ -1 +1,2 @@ | |||
| from .interpretable_predictor import InterpretableTabularPredictor | |||
| from .predictor import TabularPredictor | |||
There was a problem hiding this comment.
nit: empty line at the EOF missing
| def fit(self, | ||
| train_data, | ||
| tuning_data=None, | ||
| time_limit=None, | ||
| *, | ||
| presets='interpretable', | ||
| fit_weighted_ensemble=False, | ||
| calibrate=False, | ||
| **kwargs): |
There was a problem hiding this comment.
nit: data types; add docstrings
There was a problem hiding this comment.
keeping to align with TabularPredictor, we can add type hints to TabularPredictor later on, but the type hints are already in TabularPredictor's docstring, which this method inherits.
| for i in range(summaries.shape[0]): | ||
| model_name = summaries.index.values[i] | ||
| complexities.append(info['model_info'][model_name].get('complexity', np.nan)) | ||
| summaries.insert(2, 'complexity', complexities) |
There was a problem hiding this comment.
why insert at specific position (2)? magic number needs an explanation
| model_name = summaries_filtered.index.values[0] # best model is at top | ||
| agmodel = self._trainer.load_model(model_name) | ||
| imodel = agmodel.model | ||
| print(imodel) |
There was a problem hiding this comment.
Use logging; print won't get into a file if filehandler is used
There was a problem hiding this comment.
keeping as print, we use print for leaderboard, and this is similar
Innixma
force-pushed
the
cleanup_interpret
branch
from
52105ae
to
f4481c7
Compare
Innixma
force-pushed
the
cleanup_interpret
branch
from
f4481c7
to
2ffdf82
Compare
|
Job PR-2981-2ffdf82 is done. |
Issue #, if available:
Description of changes:
Moved interpretable logic to its own predictor class
Removed non-interpretable models from the trained models in interpretable setting
Removed weighted ensemble in interpretable setting
Disabled calibration in interpretable setting
Added experimental warnings to interpretable setting and tutorial
DO NOT MERGE until new website stable is live, this code represents a breaking change in a tutorial
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.