"""Par instant : capteurs dont corr s'écarte de leur médiane glissante (±1 j) de plus de K bruits, rejets, error_value des références.""" import csv, glob, statistics as st D="data/BREITHORN"; K=5 rows={} for f in sorted(glob.glob(f"{D}/displacement/*.csv")): for r in csv.DictReader(open(f)): k=r["time_utc"][:16].replace(" ","T") if "2026-10-03"<=k<="2026-11-01T00:01": rows.setdefault(r["sensor_name"],{})[k]=r inst=sorted({k for s in rows.values() for k in s}) ref={s for s,d in rows.items() if next(iter(d.values()))["pattern_type"]=="global_ref"} res={} for s,d in rows.items(): for c in ("corr_x","corr_y"): ser=[(k,float(d[k][c])) if k in d and d[k][c] and float(d[k]["error_value"])<=float(d[k]["error_threshold"]) else (k,None) for k in inst] vals=[v for _,v in ser] noise=st.median(abs(vals[i+1]-vals[i]) for i in range(len(vals)-1) if vals[i] is not None and vals[i+1] is not None) for i,(k,v) in enumerate(ser): if v is None: continue w=[x for x in vals[max(0,i-6):i+7] if x is not None] dv=v-st.median(w) if abs(dv)>K*noise: res.setdefault(k,[]).append(f"{s}.{c[-1]}{dv/noise:+.0f}") for k in inst: rej=[s for s in rows if k in rows[s] and float(rows[s][k]["error_value"])>float(rows[s][k]["error_threshold"])] re=" ".join(f"{s}:{float(rows[s][k]['error_value']):.2f}/{rows[s][k]['inlier'] or '-'}" for s in sorted(ref) if k in rows[s]) sg=next(iter(rows.values()))[k]["sigma0"] if k in next(iter(rows.values())) else "" if res.get(k) or rej: print(k,"| rej:",",".join(rej) or "-","| corr:"," ".join(res.get(k,[])),"| refs:",re,"| sigma0",sg[:7])