Getting Prediction Error: AttributeError: 'StatsForecast' object has no attribute 'fitted_'

[guzinerdem]

Hi,
After training and saving models with statsforecast, when I try to make prediction with sf.predict() I get AttributeError: 'StatsForecast' object has no attribute 'fitted_', how can I make prediction easily?

[jmoralez]

Hey @guzinerdem, thanks for using statsforecast. Can you please provide the code that you're running?

[Lineker98]

I am getting the same error, following the instructions to use Mlflow, it is not pissible to fit the model, when a call sf.fit(), i get 'StatsForecast' object has no attribute "'ga'". And when i log the model without fit, and trying to predict, i get 'StatsForecast' object has no attribute 'fitted_'

AttributeError: 'StatsForecast' object has no attribute 'fitted_'

AttributeError Traceback (most recent call last)
File :1
----> 1 results = loaded_model.predict(h=HORIZON, X_df=X_test, level=LEVEL)

File /local_disk0/.ephemeral_nfs/cluster_libraries/python/lib/python3.9/site-packages/statsforecast/core.py:976, in _StatsForecast.predict(self, h, X_df, level)
974 X, level = self.parse_X_level(h=h, X=X_df, level=level)
975 if self.n_jobs == 1:
--> 976 fcsts, cols = self.ga.predict(fm=self.fitted, h=h, X=X, level=level)
977 else:
978 fcsts, cols = self._predict_parallel(h=h, X=X, level=level)

AttributeError: 'StatsForecast' object has no attribute 'fitted_'

[jmoralez]

Are you following this guide? I'm able to run it end to end (except I just need to change the import from import mlflavors to import mlflavors.statsforecast)