| { |
| "analysis": "Post-hoc descriptive recount of saved step3368 validation predictions.", |
| "reporting_change_date": "2026-11-06", |
| "threshold_selection": "User selected 50/1000 after inspecting the original results; 20/1000 retained for comparison.", |
| "prediction_file_sha256": "7dad166f6d93202a410b91da3cae7db936c2d9ab6835282ad7f5a8c8b0bca9c5", |
| "checkpoint_sha256": "bd82d3c508a7b591b7c1578614231b142689982c842c6864c80baaa213e5dde5", |
| "population": 1967, |
| "denominator": "All reference actions with a coordinate field, including missing/invalid predictions as non-matches.", |
| "criterion": "Action type equality AND abs(predicted_x-reference_x) <= threshold AND abs(predicted_y-reference_y) <= threshold.", |
| "coordinate_scale": "0-1000 per axis; 50 is 5% of screen width horizontally and 5% of screen height vertically.", |
| "scope": "Primary coordinate only; no widget-boundary, drag-endpoint, scroll-amount, other-parameter or environment-success check.", |
| "inference_rerun": false, |
| "environment_rerun": false, |
| "results": [ |
| { |
| "threshold_per_axis_0_1000": 20, |
| "matched": 456, |
| "total": 1104, |
| "rate": 0.41304347826086957 |
| }, |
| { |
| "threshold_per_axis_0_1000": 50, |
| "matched": 548, |
| "total": 1104, |
| "rate": 0.4963768115942029 |
| } |
| ] |
| } |
|
|
