Model card

Name and version: __ Date: _ Authors: ___

• Purpose and intended use (the decision it informs; who uses it):
• Not intended for (uses to avoid):
• Data: source, licence, period, size, how cleaned, who is missing:
• Method: baseline, model(s), features, validation approach:
• Performance: baseline vs model on held-out test data (accuracy, precision, recall as relevant); threshold chosen and why:
• Group audit: performance and selection rate by group; group sizes; gaps found:
• Limits and risks: what could go wrong for real people; how errors are handled; humans in the loop:
• Reproduce: environment versions, random seed, steps:
• AI assistance used for:
• Contact: