Jigyasu Krishnan
Jigyasu Krishnan
> Nice, it looks like the problem with `apply_map` is solved! > > The current problem looks different - looking at the error logs, it seems some logic from forecasting...
> retrieves the defaults if scoring=None, for estimator type added this logic the tests will currently fail, because `_check_scores` returns a list and `_evaluate_fold` expects a dict, since earlier it...
@fkiraly apologies, I already detailed it, but I did it on Discord The existing evaluate function for TSF uses metrics from sktime, and those metrics are estimators with tags. The...
I have updated the `_check_scores` to be compatible for time series classification with sklearn metrics. Following @fkiraly's suggestions, the function now segragates metrics suited for probabilistic and deterministic function with...
I have removed all the logic related to `cutoff`. Currently the tests will fail because `parallelize` returns an empty structure, `results` before getting passed into `parallelize` contains the evaluate results....
Oops, I didn't realise that the result was not displaying `y_pred` and `pred_time` (learning for me to add more robust tests), made some changes to enable that but now it...
> estimator is not getting fitted fixed
I have currently hardcoded positive label to be 1 for metrics which expects a `pos_label`. Please let me know if there is any better design.
`SktimeForecastingExperiment` is WIP.
Hi @j-bowhay, thanks for the comment, I am a first-time contributor to scipy, I would like to start from this issue