Closed tulbureandreit closed 2 years ago
Hi @tulbureandreit
Thank you for bringing this to our attention.
The documentation was not quite clear on this (I just fixed it in #746), but you can actually specify a custom list of metrics.
While MSE is not a default, it can be included like this:
m = NeuralProphet(collect_metrics=["MSE", "MAE", "RMSE"])
I hope this helps?
Yes, it helps. thanks!
Hi,
When I want to be able to use the MSE in metrics when getting training and evaluation data, I am not able to use MSE although MSE is stated in the documentation as being something that is available.
Moreover, I looked into the forecaster.py file and -> and in line 295, in the collect metrics list "mse" is omitted.
Hence, if you add in that list also mse, everything works fine.
To Reproduce
Use any kind of data and when collecting the METRICS, use ""MSE""
Expected behavior The MSE shall be returned
What actually happens KeyError: "['MSE'] not in index"
Environement (please complete the following information):
pip install neuralprophet
]: last version of np installed via PyPi: neuralprophet 0.3.2Additional context
Please see the what happened for the fix. Add mse in forecaster .py in line 295 in the metrics list.
Thank you.