extending gluonts deepVAR to include imputation for all variables #2826
Unanswered
ShirleyMgit
asked this question in
Q&A
Replies: 1 comment
-
understood the problem. solved |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Dear team,
I have extended the deepVAR to include imputation for all dimensions (by treating each dimension as univariate and ignoring correlation between them), further I have added static features (similarly to deepAR). I can train the network but when trying to upload the predictor network I get an error.
For the multivariate imputation I have added a class within the network code and it seems that the predictor class does not recognize it
I get the following error:
AssertionError: Can not locate main.RollingMultivariateMeanValueImputation. (this is the new imputation)
the imputation is done within create_transformation:
AddObservedValuesIndicator(
target_field=FieldName.TARGET,
output_field=FieldName.OBSERVED_VALUES,
imputation_method=RollingMultivariateMeanValueImputation(window_size=self.impuation_window_size)
),
and it falls here:
le ~/anaconda3/envs/python3/lib/python3.10/site-packages/gluonts/model/predictor.py:118, in Predictor.deserialize(cls, path, **kwargs)
112 raise OSError(
113 f"Class {fqname_for(tpe)} is not "
114 f"a subclass of {fqname_for(Predictor)}"
115 )
117 # call deserialize() for the concrete Predictor type
--> 118 return tpe.deserialize(path, **kwargs)
which seems that it does not recognize the predictor as a subtype of Predictor.
what should I do?
thanks
Beta Was this translation helpful? Give feedback.
All reactions