File size: 3,136 Bytes
2e511b5 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 | import csv
from legex.analysis.iaa import (
PRIMARY,
cohen_kappa,
pairwise_agreement,
summarise_by_field,
write_kappa_audit_csv,
)
def test_cohen_kappa_perfect_agreement() -> None:
pairs = [("a", "a"), ("b", "b"), ("a", "a"), ("b", "b")]
assert cohen_kappa(pairs) == 1.0
def test_cohen_kappa_chance_agreement_is_zero() -> None:
# Both annotators split 50/50 but never coincide better than chance.
pairs = [("a", "a"), ("a", "b"), ("b", "a"), ("b", "b")]
k = cohen_kappa(pairs)
assert k is not None
assert abs(k) < 1e-9
def test_cohen_kappa_single_category_undefined() -> None:
assert cohen_kappa([("a", "a"), ("a", "a")]) is None
def test_cohen_kappa_too_few_items() -> None:
assert cohen_kappa([("a", "b")]) is None
def _labels_two_annotators():
# Two annotators, one country, three cases, two fields.
return {
(PRIMARY, "ch", "c1"): {"f1": "x", "f2": "1"},
(PRIMARY, "ch", "c2"): {"f1": "y", "f2": "2"},
(PRIMARY, "ch", "c3"): {"f1": "z", "f2": ""},
("de", "ch", "c1"): {"f1": "x", "f2": "1"},
("de", "ch", "c2"): {"f1": "WRONG", "f2": "2"},
("de", "ch", "c3"): {"f1": "z", "f2": ""},
}
def test_pairwise_agreement_counts() -> None:
rows = pairwise_agreement(_labels_two_annotators())
by_field = {r.field: r for r in rows}
assert by_field["f1"].n == 3
assert by_field["f1"].n_agree_exact == 2 # c1, c3 agree; c2 differs
assert by_field["f2"].n_agree_exact == 3
assert by_field["f2"].pct_exact == 1.0
def test_pairwise_tolerant_handles_numeric() -> None:
labels = {
(PRIMARY, "ch", "c1"): {"amount": "1500"},
("de", "ch", "c1"): {"amount": "1500.0"},
}
rows = pairwise_agreement(labels)
assert rows[0].n_agree_tolerant == 1
def test_summarise_by_field_weights_by_n() -> None:
rows = pairwise_agreement(_labels_two_annotators())
summary = summarise_by_field(rows)
assert summary["f2"]["pct_exact"] == 1.0
assert 0.0 <= summary["f1"]["pct_exact"] <= 1.0
def test_write_kappa_audit_csv(tmp_path) -> None:
labels = {
(PRIMARY, "ch", "c1"): {"f1": "x", "amount": "1500"},
(PRIMARY, "ch", "c2"): {"f1": "y", "amount": "10"},
("de", "ch", "c1"): {"f1": "x", "amount": "1500.0"},
("de", "ch", "c2"): {"f1": "WRONG", "amount": "10"},
}
path = tmp_path / "kappa_audit.csv"
n = write_kappa_audit_csv(labels, path)
rows = list(csv.DictReader(path.open(encoding="utf-8")))
# 1 country, 1 pair, 2 shared cases, 2 fields -> 4 cells.
assert n == 4 == len(rows)
cell = {(r["case_id"], r["field"]): r for r in rows}
# f1 on c1 agrees both ways; amount 1500 vs 1500.0 is tolerant-only.
assert cell[("c1", "f1")]["decision_exact"] == "yes"
assert cell[("c1", "f1")]["decision_tolerant"] == "yes"
assert cell[("c1", "amount")]["decision_exact"] == "no"
assert cell[("c1", "amount")]["decision_tolerant"] == "yes"
# f1 on c2 disagrees both ways.
assert cell[("c2", "f1")]["decision_exact"] == "no"
assert cell[("c2", "f1")]["decision_tolerant"] == "no"
|