SICSS 2026 Synthetic Shoppers Challenge · Public leaderboard
SICSS 2026 — Leaderboard
Test-set scores for submitted predictions. Lower is better for all metrics.
How ranking works
- Primary sort: overall MAE (mean absolute error vs. hidden test probabilities)
- Persona columns: MAE within each shopper segment (Value, Trust, Convenience, Inspiration)
- Brier: squared probability error — rewards calibration
- Design-effect MAE: error in A→B lift per segment×category pair
- Each submission is shown separately