Module 3.4.3 · Phase 3 Intervention · Step 3.4 TechniquesPreferred
Post-Processing Fairness Techniques
Can the fix happen in the model's outputs?
Adjusts the scores, probabilities or decisions of a trained model to meet a fairness goal when retraining is not possible.
What it does
You choose and document threshold, calibration, score transformation or reject-option techniques and plan their validation. Supported techniques run on the predictions in your dataset and show fairness metrics before and after.
Why it matters
This step selects and configures specific pre-processing, in-processing, or post-processing techniques that implement the approved intervention plan. Each technique choice is documented with causal justification, configuration rationale, and expected fairness impact.
Where it sits
- Causal Model
- Counterfactual Test
- Intervention Plan
- Techniques
- Architecture Guidance
- Pre-Processing Fairness Techniques
- In-Processing Fairness Techniques
- Post-Processing Fairness Techniques
How to use it
In the Fairness Navigator Beta
Work through the steps in order. Several modules open from the Navigator's steps and hand their results back.
On its own Live
Open 3.4.3 from the platform's Modules page and run it when your project needs it. Work is saved with the assessment.
After a Pulse reading Live
Start with a fast first read, then go deeper with the modules where it points.