Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

&yarnabrina [BUG] correct ForecastX behaviour in case of multivariate y #4719

Merged
merged 2 commits into from Jun 20, 2023

Conversation

fkiraly
Copy link
Collaborator

@fkiraly fkiraly commented Jun 19, 2023

For multivariate y, the compositor ForecastX would always broadcast over variables of y, even if wrapped forecasters were multivariate.

However, ForecastX itself should never broadcast over y, that should be left to component forecasters.

Fixed here, this also fixes #4718

Includes a slightly extended test case from the bug report #4718 by @yarnabrina

@fkiraly fkiraly added module:forecasting forecasting module: forecasting, incl probabilistic and hierarchical forecasting bugfix Fixes a known bug or removes unintended behavior labels Jun 19, 2023
@fkiraly fkiraly changed the title [BUG] correct ForecastX behaviour in case of multivariate y &yarnabrina [BUG] correct ForecastX behaviour in case of multivariate y Jun 19, 2023
@yarnabrina
Copy link
Collaborator

The error in macOS runner with importing pandas:

ModuleNotFoundError: No module named '_bz2'

I suppose this can be ignored? Only happened on this specific job, others ran fine.

@fkiraly
Copy link
Collaborator Author

fkiraly commented Jun 19, 2023

I suppose this can be ignored? Only happened on this specific job, others ran fine.

Not sure - could be a 3.7 issue.

@fkiraly fkiraly merged commit d9dc6d3 into main Jun 20, 2023
27 of 28 checks passed
@fkiraly fkiraly deleted the forecastX-multivariate branch June 20, 2023 22:20
CTFallon pushed a commit to CTFallon/sktime that referenced this pull request Jun 21, 2023
…te `y` (sktime#4719)

For multivariate `y`, the compositor `ForecastX` would always broadcast
over variables of `y`, even if wrapped forecasters were multivariate.

However, `ForecastX` itself should never broadcast over `y`, that should
be left to component forecasters.

Fixed here, this also fixes sktime#4718

Includes a slightly extended test case from the bug report sktime#4718 by
@yarnabrina
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bugfix Fixes a known bug or removes unintended behavior module:forecasting forecasting module: forecasting, incl probabilistic and hierarchical forecasting
Projects
None yet
Development

Successfully merging this pull request may close these issues.

[BUG] missing forecaster_X_ attribute in ForecastX estimator in case of multivariate target
2 participants