Skip to content

Add FLAIR statistical method results#120

Open
TakatoHonda wants to merge 1 commit intoSalesforceAIResearch:mainfrom
TakatoHonda:add-flair
Open

Add FLAIR statistical method results#120
TakatoHonda wants to merge 1 commit intoSalesforceAIResearch:mainfrom
TakatoHonda:add-flair

Conversation

@TakatoHonda
Copy link

Summary

Add results for FLAIR (Factored Level And Interleaved Ridge), a statistical forecasting method.

  • Model type: statistical (Ridge regression, no neural network)
  • Organization: Mellon Inc.
  • Test data leakage: No

Method

FLAIR separates what (level) from how (shape) via period-aligned matrix reshaping:

y(phase, period) = Level(period) × Shape(phase)

Level is forecast by Ridge regression with soft-average GCV (one SVD). Shape is estimated structurally from recent within-period proportions.

Results

  • 97/97 configurations evaluated
  • relMASE: 1.028 (relative to Seasonal Naive)
  • relCRPS: 0.750

@salesforce-cla
Copy link

Thanks for the contribution! Before we can merge this, we need @TakatoHonda to sign the Salesforce Inc. Contributor License Agreement.

@TakatoHonda
Copy link
Author

I have read the CLA Document and I hereby sign the CLA. @salesforce-cla

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant