Human Calibration
Compares Customer Twin prediction against real customer evaluation.
-0.06
Spearman ρ
Rank alignment across configurations
0.36
MAE
Average absolute score difference
55%
Decision Agreement
BUY/CONSIDER/REJECT agreement
0.91
Stability
1 − normalized MAE
AI–Human Alignment
Illustrative calibration simulation
Calibration Metrics
| Metric | Value | Interpretation |
|---|---|---|
| Spearman ρ | -0.06 | Rank alignment across configurations |
| MAE | 0.36 | Average absolute score difference (1-5 scale) |
| Decision Agreement | 55% | BUY/CONSIDER/REJECT agreement |
| Stability | 0.91 | 1 − normalized MAE across illustrative responses |
These values demonstrate how the system would report calibration after real human validation data are collected. Computed from 14 configuration(s) with ≥10 respondents each.