forked from chiki2bum2/SniperGold_ML
21 lines
840 B
Python
21 lines
840 B
Python
# -*- coding: utf-8 -*-
| |||
"""Review cepat hasil machine annotation (sanity check)."""
| |||
import csv
| |||
import json
| |||
| |||
rows = list(csv.DictReader(open("output/machine_annotations.csv", encoding="utf-8-sig")))
| |||
print("n=", len(rows))
| |||
for r in rows[:6]:
| |||
print(r["case_id"], r["decision_timestamp"], r["machine_decision"],
| |||
r["primary_primitive"], "|", r["reference_type"],
| |||
"| rej:", r["rejection"][:40], "|", r["machine_reason"][:110])
| |||
print("...")
| |||
for r in rows[-4:]:
| |||
print(r["case_id"], r["decision_timestamp"], r["machine_decision"],
| |||
r["primary_primitive"], "|", r["reference_type"],
| |||
"|", r["machine_reason"][:110])
| |||
| |||
meta = json.load(open("output/cases_meta.json", encoding="utf-8"))
| |||
print("\nstrata di meta:")
| |||
from collections import Counter
| |||
print(Counter(c["sampling_stratum"] for c in meta["cases"]))
|