- 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>
68 lines
3.4 KiB
Python
68 lines
3.4 KiB
Python
"""Requalification: per label, the largest deviation of each target in its own noises.
|
|
Noise: median |step| between consecutive accepted measurements over the analysis window (as ranges.py).
|
|
Levels: 'win' = median over the analysis window (ranges.py default); 'roll' = median of +-4 neighbours
|
|
without the point (ranges.py --rolling 4); 'night' (thermal labels) = median of 01:00-05:00 UTC of the
|
|
label's start day. Prints the top sensors per label, and the rejected count of each target within it.
|
|
Usage: python3 requalify.py labels.jsonl"""
|
|
import sys, json, statistics
|
|
sys.path.insert(0, 'tools/scripts')
|
|
import data
|
|
from datetime import timedelta
|
|
|
|
S, E = data.parse_time('2026-09-17T05:45Z'), data.parse_time('2026-09-30T04:15Z')
|
|
by = {s: data.filter_window(m, S - timedelta(hours=6), E) for s, m in data.load('data/NORDEND').items()}
|
|
refs = {s for s, ms in by.items() if data.is_reference(ms)}
|
|
acc = {s: [m for m in ms if not data.rejected(m)] for s, ms in by.items()}
|
|
noise, level = {}, {}
|
|
for s, ms in acc.items():
|
|
for c in ('corr_x', 'corr_y'):
|
|
v = [m[c] for m in ms if m[c] is not None and S <= m['t'] <= E]
|
|
d = [abs(v[i + 1] - v[i]) for i in range(len(v) - 1)]
|
|
noise[s, c] = statistics.median(d) if d else None
|
|
level[s, c] = statistics.median(v) if v else None
|
|
|
|
def dev(lab):
|
|
a, b = data.parse_time(lab['start']), data.parse_time(lab['end'])
|
|
night = None
|
|
if lab['family'] == 'Thermal displacement':
|
|
n0 = a.replace(hour=1, minute=0); n1 = a.replace(hour=5, minute=0)
|
|
night = (n0, n1)
|
|
rows = []
|
|
for s, ms in acc.items():
|
|
if s in refs: continue
|
|
for c in ('corr_x', 'corr_y'):
|
|
nz = noise[s, c]
|
|
if not nz: continue
|
|
vals = [(i, m) for i, m in enumerate(ms) if a <= m['t'] <= b and m[c] is not None]
|
|
if not vals: continue
|
|
res = {}
|
|
res['win'] = max(((m[c] - level[s, c]) / nz for i, m in vals), key=abs)
|
|
rr = []
|
|
for i, m in vals:
|
|
nb = [ms[j][c] for j in range(max(0, i - 4), min(len(ms), i + 5)) if j != i and ms[j][c] is not None]
|
|
if len(nb) >= 3: rr.append(((m[c] - statistics.median(nb)) / nz, m['t']))
|
|
if rr: res['roll'] = max(rr, key=lambda x: abs(x[0]))
|
|
if night:
|
|
nv = [m[c] for m in ms if night[0] <= m['t'] <= night[1] and m[c] is not None]
|
|
if len(nv) >= 5:
|
|
lv = statistics.median(nv)
|
|
res['night'] = max((((m[c] - lv) / nz, m['t']) for i, m in vals), key=lambda x: abs(x[0]))
|
|
rows.append((s, c, res))
|
|
rej = {s: sum(1 for m in by[s] if a <= m['t'] <= b and data.rejected(m)) for s in by}
|
|
return rows, rej
|
|
|
|
for line in open(sys.argv[1]):
|
|
lab = json.loads(line)
|
|
rows, rej = dev(lab)
|
|
print(f"\n== {lab.get('id','')} {lab['start']} -> {lab['end']} {lab['family']}")
|
|
for key in ('night', 'roll', 'win'):
|
|
cand = [(s, c, r[key]) for s, c, r in rows if key in r]
|
|
if not cand: continue
|
|
val = lambda x: x[2][0] if isinstance(x[2], tuple) else x[2]
|
|
cand.sort(key=lambda x: -abs(val(x)))
|
|
txt = ', '.join(f"{s}.{c[-1]} {val(x):+.1f}" + (f"@{data.iso(x[2][1])[5:16]}" if isinstance(x[2], tuple) else '')
|
|
for x in cand[:6] for s, c in [(x[0], x[1])])
|
|
print(f" {key:5s}: {txt}")
|
|
r = {s: n for s, n in rej.items() if n}
|
|
print(' rejected:', r)
|