Files
aimsight-labeling/auto/analyses/WILDSTRUBEL_2026-09-01T0000_2026-09-13T0000/scripts/check_labels.py
T
maximeandClaude Opus 5.5 ddb17c2189 auto: labels requalified under the 5-noise rules (only what stands out from the noise): 32 analyses refined, the 6 paused periods verified, the long labels requalified with the fixes the period analysts found; 1153 -> 961 labels
- 32 analyses: each label kept, rewritten or removed by its analyst (section Requalification of the report; mm magnitudes where the targets have them)
- SNCF01 x3, WILDSTRUBEL 2026-09-01 -> 2026-10-06 x3: verified from their existing labels under the new rules
- long labels (no analysis): auto/analyses/long-labels/<STATION>/; NORDEND-169 and -195 run to 07/11, so NORDEND-283 and -284 (their continuation) are removed
- reading-rules.md, stations.md: the citations of removed or rewritten labels point to their successor, or to [expert]

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-17 23:58:00 +01:00

84 lines
3.2 KiB
Python

"""For each label of a jsonl: per target, the largest deviation of corr_x, corr_y (accepted measurements) inside
[start, end], in own noises, against (a) the window level (median), (b) the level of the preceding night 20-03 UTC,
(c) for short labels (<= 4 steps), the jump against the mean of the measurement before and after the label.
Usage: python3 check_labels.py <labels.jsonl> [--sensors A,B]"""
import json
import statistics
import sys
from datetime import timedelta
sys.path.insert(0, "tools/scripts")
import data # noqa: E402
import common # noqa: E402
FOLDER = "data/WILDSTRUBEL"
W0 = data.parse_time("2026-09-01T00:00Z")
W1 = data.parse_time("2026-09-09T21:45Z")
by = data.load(FOLDER)
targets = [s for s, ms in by.items() if not data.is_reference(ms)]
series = {}
noise = {}
for s in targets:
ms = [m for m in by[s] if W0 - timedelta(days=1) <= m["t"] <= W1]
for c in ("corr_x", "corr_y"):
pts = [(m["t"].replace(second=0, microsecond=0), m[c]) for m in ms if m[c] is not None]
series[(s, c)] = pts
noise[(s, c)] = common.own_noise([v for t, v in pts if t >= W0])
def night_level(s, c, t):
d = t - timedelta(hours=t.hour, minutes=t.minute)
if t.hour < 20:
d -= timedelta(days=1)
a, b = d + timedelta(hours=20), d + timedelta(hours=27)
v = [x for tt, x in series[(s, c)] if a <= tt <= b]
return statistics.median(v) if v else None
def report(lab):
a, b = data.parse_time(lab["start"]), data.parse_time(lab["end"])
rows = []
for s in targets:
for c in ("corr_x", "corr_y"):
pts = series[(s, c)]
n = noise[(s, c)]
inside = [(t, v) for t, v in pts if a < t < b] or [(t, v) for t, v in pts if a <= t <= b]
if not inside or not n:
continue
lvl = night_level(s, c, a)
dev_n = max(((v - lvl) / n, t) for t, v in inside) if lvl is not None else None
dev_n2 = min(((v - lvl) / n, t) for t, v in inside) if lvl is not None else None
jump = None
before = [v for t, v in pts if t <= a][-1:]
after = [v for t, v in pts if t >= b][:1]
if before and after and (b - a) <= timedelta(minutes=75):
ref = (before[0] + after[0]) / 2
jump = max(((v - ref) / n, t) for t, v in inside), min(((v - ref) / n, t) for t, v in inside)
rows.append((s, c, dev_n, dev_n2, jump))
return rows
def main():
path = sys.argv[1]
for line in open(path):
lab = json.loads(line)
print(f"== {lab.get('id','')} {lab['start']} {lab['end']} {lab['family']}")
for s, c, hi, lo, jump in report(lab):
parts = []
if jump:
(jh, th), (jl, tl) = jump
j = jh if abs(jh) >= abs(jl) else jl
tj = th if abs(jh) >= abs(jl) else tl
if abs(j) >= 3:
parts.append(f"jump {j:+.1f}n @{tj:%d %H:%M}")
if hi and lo:
d, t = hi if abs(hi[0]) >= abs(lo[0]) else lo
if abs(d) >= 5 and not jump:
parts.append(f"vs night {d:+.1f}n @{t:%d %H:%M}")
if parts:
print(f" {s:6} {c}: " + "; ".join(parts))
if __name__ == "__main__":
main()