-
Notifications
You must be signed in to change notification settings - Fork 525
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
Raise ARIMA parameter limits from 4 to 8 #4022
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks Louis for this PR, it looks good to me.
Do I understand correctly, that one could (in theory) use the jones_transform_kernel
with higher VALUE
template parameter? So setting the value to 8 is not limited by the kernel resource usage, instead it is chosen so because higher values are not useful in practice?
Could you update the PR description to clarify this?
It's a bit of both. With the current implementation using a template parameter, we have to choose an arbitrary limit, and 8 should be more than enough (in practise one rarely needs more than 3 or 4) |
The HDBSCAN error found in CI I think is fixed now, so rerunning tests |
rerun tests |
@gpucibot merge |
Codecov Report
@@ Coverage Diff @@
## branch-21.08 #4022 +/- ##
===============================================
Coverage ? 85.59%
===============================================
Files ? 230
Lines ? 18349
Branches ? 0
===============================================
Hits ? 15705
Misses ? 2644
Partials ? 0
Flags with carried forward coverage won't be shown. Click here to find out more. Continue to review full report at Codecov.
|
Resolves rapidsai#3915 This PR still doesn't allow arbitrary parameters (because of the Jones transform kernel that uses the parameters as a template argument in order to work with registers). But fixing the limit to 8 instead of 4 should be enough. In most cases, the parameters don't exceed 3 or 4, and more complex models don't give better results. I also added tests with `p=5`. Authors: - Louis Sugy (https://github.com/Nyrio) Approvers: - Tamas Bela Feher (https://github.com/tfeher) - Dante Gama Dessavre (https://github.com/dantegd) URL: rapidsai#4022
Resolves #3915
This PR still doesn't allow arbitrary parameters (because of the Jones transform kernel that uses the parameters as a template argument in order to work with registers).
But fixing the limit to 8 instead of 4 should be enough. In most cases, the parameters don't exceed 3 or 4, and more complex models don't give better results.
I also added tests with
p=5
.