[ENH] k-nearest neighbors regressor: support for non-brute algorithms and non-precomputed mode to improve memory efficiency

[Z-Fran]

Reference Issues/PRs

#3429 (comment)

What does this implement/fix? Explain your changes.

adds test parameter sets for KNeighborsTimeSeriesRegressor;

adds support for non-brute algorithms and non-precomputed mode, mirroring #5937

For all contributions

• I've added myself to the list of contributors with any new badges I've earned :-)
  How to: add yourself to the all-contributors file in the sktime root directory (not the CONTRIBUTORS.md). Common badges: code - fixing a bug, or adding code logic. doc - writing or improving documentation or docstrings. bug - reporting or diagnosing a bug (get this plus code if you also fixed the bug in the PR).maintenance - CI, test framework, release.
  See here for full badge reference
• Optionally, for added estimators: I've added myself and possibly to the maintainers tag - do this if you want to become the owner or maintainer of an estimator you added.
  See here for further details on the algorithm maintainer role.
• The PR title starts with either [ENH], [MNT], [DOC], or [BUG]. [BUG] - bugfix, [MNT] - CI, test framework, [ENH] - adding or improving code, [DOC] - writing or improving documentation or docstrings.

[yarnabrina]

Thanks for your contribution. I think your changes are fine on their own, but you probably discovered some previously unknown bugs that's causing failures in CI.

@fkiraly can you please take a look? This seems to be coming from this line:

sktime/sktime/regression/distance_based/_time_series_neighbors.py

Line 134 in e0e66f1

metric="precomputed",

[Z-Fran]

Hi @yarnabrina @fkiraly. It seems like sklearn doesn't support all tree algorithms to accept a precomputed distance matrix. (mentioned there)
And I find that #5937 fixes the same bug of KNeighborsTimeSeriesClassifier. Can I fix KNeighborsTimeSeriesRegressor just like it?

BTW, could you please tell me why the parameter weights is not used in KNeighborsTimeSeriesRegressor but used in KNeighborsTimeSeriesClassifier (added in #1998)

[fkiraly]

@Z-Fran, thanks!

Well-spotted, indeed we find bugs in a couple places when we add new test parameters.
But you've already identified and fixed it, nice!

I'm happy to merge this PR, although it introduces duplication in the kneighbors estimators.

Generally, if you end up copying multiple lines of code, that is an "antipattern", violating the DRY ("don't repeat yourself") principle.
Simlar code also exists in other places, in relation to the distances, so it might be good to move it into a single location, e.g., in a new module distances._adapters.sklearn.

I'd consider that a task for a separate PR though - so you can take a stab at it, or we should track it as an issue for someone else to pick up ("resolve duplication in...")

More generally, there's some open work around the distances, which could be a nice longer project if you are interested.

[Z-Fran]

@Z-Fran, thanks!

Well-spotted, indeed we find bugs in a couple places when we add new test parameters. But you've already identified and fixed it, nice!

I'm happy to merge this PR, although it introduces duplication in the kneighbors estimators.

Generally, if you end up copying multiple lines of code, that is an "antipattern", violating the DRY ("don't repeat yourself") principle. Simlar code also exists in other places, in relation to the distances, so it might be good to move it into a single location, e.g., in a new module distances._adapters.sklearn.

I'd consider that a task for a separate PR though - so you can take a stab at it, or we should track it as an issue for someone else to pick up ("resolve duplication in...")

More generally, there's some open work around the distances, which could be a nice longer project if you are interested.

@fkiraly I'm sorry for not getting back to you sooner. I will resolve the duplication for a separate PR after merging this PR. I will also track some open work around the distances in my free time.