Step 2: the evaluator (the measure of bench/score.py on predictions per instant; the rules reproduce it exactly)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
maximeandClaude Opus 5.5 committed 2026-11-16 16:25:03 +01:00
1 parent b2d3734152
commit c04cbc90bb
4 files changed
+153 -1

No files matched your search

+18
View File
@@ -0,0 +1,18 @@
"""Checks of evaluate.py: python test_evaluate.py (the check against the benchmark itself is `evaluate.py rules`)."""
import numpy as np
import evaluate
instants = ["2026-10-01T00:00Z", "2026-10-01T00:15Z", "2026-10-01T00:30Z"]
yhat = evaluate.events_to_yhat([{"start": "2026-10-01T00:15Z", "end": "2026-10-01T00:30Z", "category": "Obstruction"},
{"start": "2026-10-01T00:00Z", "end": "2026-10-01T00:00Z", "kind": "Visibility"}],
instants)
assert yhat[:, 1].tolist() == [0, 1, 1] and yhat[:, 0].tolist() == [1, 0, 0] # bounds included
y = np.zeros_like(yhat)
y[1:, 1] = 1
y[2, 0] = 1
c = evaluate.counts(y, yhat)
assert c == {"Visibility": [1, 1, 0], "Obstruction": [2, 2, 2]}, c
assert evaluate.regressions({"Obstruction": [10, 10, 5]}, {"Obstruction": [10, 4, 4]}) == [("Obstruction", "found", 50, 40)]
assert evaluate.regressions({"Obstruction": [10, 10, 5]}, {"Obstruction": [10, 10, 5]}) == []
print("ok")