Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
29 changes: 29 additions & 0 deletions scripts/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -195,6 +195,35 @@ manuscript_assets/tables/figure5_fps_spearman_r2_leaderboard.tsv
manuscript_assets/tables/figure5_fps_local_ranks.tsv
```

## Generate Figure 6 FPS Recovery

Run from the repository root:

```bash
uv run python scripts/figure6_fps_recovery.py \
--run-dir results/<results_dir>
```

Figure 6 evaluates practical recovery under the Full Pair Set (FPS), using the
same rank-zero convention for missing predictor scores as Figure 5. It reports
positive-pair recovery by prediction budget, the best per-experiment precision
at a fixed recall target, and the corresponding false positives per true
positive.

This writes:

```text
manuscript_assets/figure6/figure6_fps_recovery_panel_a_recovery_budget.png
manuscript_assets/figure6/figure6_fps_recovery_panel_b_best_precision.png
manuscript_assets/figure6/figure6_fps_recovery_panel_c_validation_burden.png
manuscript_assets/figure6/figure6_fps_recovery_panel_d_legend.png
manuscript_assets/figure6/figure6_fps_recovery_combined.png
manuscript_assets/figure6/figure6_fps_recovery_combined.svg
manuscript_assets/tables/figure6_fps_recovery_by_budget.tsv
manuscript_assets/tables/figure6_fps_precision_at_recall.tsv
manuscript_assets/tables/figure6_fps_best_precision_at_recall.tsv
```

## Generate Supplementary Top-Effect CDF Figure

After running the benchmark with `evaluation.write_top_prediction_cdfs: true`,
Expand Down
Loading