Model Cards¶
Model cards for temporalcv validation components, following the Mitchell et al. (2019) framework adapted for ML infrastructure.
Overview¶
Model cards provide transparent documentation of component behavior, assumptions, and limitations. Each card includes:
Component Details: Version, module, license, knowledge tier
Intended Use: Primary use cases and out-of-scope applications
Parameters: Complete parameter reference with knowledge tier tags
Assumptions: What must hold for correct behavior
Limitations: Known constraints and common misconfigurations
References: Academic sources tagged by confidence tier
Available Model Cards¶
Component |
Type |
Card |
|---|---|---|
WalkForwardCV |
Cross-validator |
|
gate_signal_verification |
Validation gate |
Knowledge Tier System¶
All claims are tagged with confidence levels:
Tier |
Meaning |
Example |
|---|---|---|
[T1] |
Academically validated |
DM test (Diebold & Mariano 1995) |
[T2] |
Empirical finding |
Block permutation default |
[T3] |
Assumption/heuristic |
α=0.05 significance level |
Using Model Cards¶
Before Deployment¶
Read the Intended Use section to verify your use case is supported
Check Assumptions against your data and pipeline
Review Limitations for known constraints
During Development¶
Use Parameters table as a quick reference
Check Common Misconfigurations to avoid pitfalls
Follow Examples for correct usage patterns
For Documentation¶
Cite References in publications
Use Knowledge Tier tags for claim confidence
Note Limitations in methodology sections
Contributing¶
To add a new model card:
Copy the template structure from existing cards
Fill in all sections with accurate information
Tag all claims with appropriate knowledge tiers
Add academic references for [T1] claims
Document empirical sources for [T2] claims
Justify [T3] heuristics explicitly
Reference¶
Mitchell, M., Wu, S., Zaldivar, A., Barnes, P., Vasserman, L., Hutchinson, B., Spitzer, E., Raji, I.D., & Gebru, T. (2019). Model Cards for Model Reporting. Proceedings of the Conference on Fairness, Accountability, and Transparency, 220-229.