Predict with a selected modle

[xinwei-sher]

The predict function uses the best model from AutoML. I'm interested to see the predictions from other models as well. Is there any way to select one of the other models and use it for prediction? Thanks.

"predict(self, X)
Computes predictions from AutoML best model."

[pplonski]

To compute predictions from selected model you need to change the model name in the best_model.txt file, then reload the AutoML and compute predictions. It is very ugly approach. I've added issue to add such functionality to select model for predictions: #423

[elcolie]

I don't have best_model.txt. Where I can find it?

[elcolie]

I think I found it. It is params.json. Then looking for best_model

[pplonski]

@elcolie you are right, it is in params.json with key best_model. Sorry for confusion. Please let me know if it works.

[xinwei-sher]

This doesn't work.

I changed the key best_model in params.json to another model, "2_Default_LightGBM", and tried this

automl = AutoML(results_path = results_path)
automl.fit(X_train, y_train)
automl.predict(X_test)

This is the message I got:
This model has already been fitted. You can use predict methods or select a new 'results_path' for a new 'fit()'.
2021-07-09 19:29:01,810 supervised.exceptions ERROR This model has not been fitted yet. Please call fit() first.

---

AutoMLException Traceback (most recent call last)
in
----> 1 automl.predict(X_test)

~/miniconda3/envs/autogluon/lib/python3.7/site-packages/supervised/automl.py in predict(self, X)
354 AutoMLException: Model has not yet been fitted.
355 """
--> 356 return self._predict(X)
357
358 def predict_proba(self, X):

~/miniconda3/envs/autogluon/lib/python3.7/site-packages/supervised/base_automl.py in _predict(self, X)
1304 def _predict(self, X):
1305
-> 1306 predictions = self._base_predict(X)
1307 # Return predictions
1308 # If classification task the result is in column 'label'

~/miniconda3/envs/autogluon/lib/python3.7/site-packages/supervised/base_automl.py in _base_predict(self, X, model)
1241 if model is None:
1242 raise AutoMLException(
-> 1243 "This model has not been fitted yet. Please call fit() first."
1244 )
1245

AutoMLException: This model has not been fitted yet. Please call fit() first.

[pplonski]

This discussion is connected with #423

I will try to fix this soon. Apologize for problems.