Fixed Pytest warnings for machine_learning/forecasting #8958
Conversation
There was a problem hiding this comment.
LGTM! The build logs no longer show the Pytest warnings, which is nice. However, this PR doesn't completely fix #8780 as there are some deeper issues that still need to be addressed: where the code actually comes from, the use of existing ML models, and whether this code actually qualifies as an "algorithm". With that said, the more immediate issues seem to have been fixed, which is good.
|
@CaedenPH, thoughts? |
There was a problem hiding this comment.
@tianyizheng02
I would be inclined to merge this pull request but also have discussions on #8780 with what to do with this file and potentially rework later.
Therefore, I think it should close my issue (#7305) but not #8780
|
@p1utoze Thanks for your contributions! |
|
@tianyizheng02 I agree. I looked into that aspect as well that the algorithms mentioned clearly don't explain about the forecasting. Well, fixing the existing errors was much needed I felt and glad to contribute. |
…#8958) * updating DIRECTORY.md * Fixed pyTest Warnings --------- Co-authored-by: github-actions <${GITHUB_ACTOR}@users.noreply.github.com>
Describe your change:
Fixes [#7305] [#8780]
The numpy Runtime warning was due to the 0 in
seasonal_ordertuple and regarding the ARIMA models. For estimation, it needs more observations to work well and the User Warning are raised because of the sample test>>> sarimax_predictor([4,2,6,8], [3,1,2,4], [2])used by doctests.Checklist: