Skip to content

Conversation

shchur
Copy link
Collaborator

@shchur shchur commented Apr 5, 2024

Description of changes:

  • Preprocess real-valued columns in known, past & static covariates for GluonTS models. We use the same logic as used by TORCH_NN model in AG-Tabular:
    • Boolean features are kept as-is
    • Skewed features (skewness > 0.99) are transformed with QuantileTransform
    • Remaining features are transformed with StandardScaler

To do:

  • Add tests

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

@shchur shchur requested a review from canerturkmen April 5, 2024 13:42
@shchur shchur mentioned this pull request Apr 5, 2024
23 tasks
@yinweisu
Copy link
Contributor

yinweisu commented Apr 5, 2024

Previous CI Run Current CI Run

Copy link
Contributor

@canerturkmen canerturkmen left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM! Thanks!

@shchur shchur force-pushed the normalize-feat-ts branch from 3e64167 to 5b38e4f Compare April 5, 2024 14:29
@yinweisu
Copy link
Contributor

yinweisu commented Apr 5, 2024

Previous CI Run Current CI Run

@yinweisu
Copy link
Contributor

yinweisu commented Apr 5, 2024

Previous CI Run Current CI Run
fonttools==4.50.0 fonttools==4.51.0
fonttools==4.50.0 fonttools==4.51.0

@shchur shchur merged commit 5eca065 into autogluon:master Apr 5, 2024
Copy link

github-actions bot commented Apr 5, 2024

Job PR-4042-acffe07 is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-4042/acffe07/index.html

@shchur shchur deleted the normalize-feat-ts branch April 6, 2024 06:25
LennartPurucker pushed a commit to LennartPurucker/autogluon that referenced this pull request Jun 1, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants