I get an error with the base line architecture for inference.
I get the error:
scaler.transform(hourly_features),
File “/home/appuser/miniconda/envs/py/lib/python3.8/site-packages/sklearn/preprocessing/_data.py”, line 791, in transform
X = self._validate_data(X, reset=False,
File “/home/appuser/miniconda/envs/py/lib/python3.8/site-packages/sklearn/base.py”, line 436, in _validate_data
self._check_n_features(X, reset=reset)
File “/home/appuser/miniconda/envs/py/lib/python3.8/site-packages/sklearn/base.py”, line 372, in check_n_features
raise RuntimeError(
RuntimeError: The reset parameter is False but there is no n_features_in attribute. Is this estimator fitted?
it seems that it is due to the scaler…I fitted the scaler (and model) in google colab.
Many Thanks in advance.
Hi!
I had the same error, and somewhere above in the error trace it was written that the version in which you save the scaler is different from the one used in the docker. What helped me to solve the error is upgrading the scikit-learn to the latest one.
Good luck!
Thank you for the fast respons!!!
I had this error one hour ago, so I feel your pain)
Hope it works for you!
I i just updated my sklearn version to 0.23.2 and upload the new scaler with my submission. Now i wait and cross fingers 
If it does not work, try following instructions that are listed here:
https://github.com/drivendataorg/noaa-runtime
I had lots of problems, so I used it to debug my submission.
It works!! Thank you very much. Now I have working pipeline this makes submitions simpler