56 lines
2.3 KiB
Python
56 lines
2.3 KiB
Python
|
"""Unit tests for the drift-check comparison logic (hermetic — no network). The
|
||
|
|
source fetch (check_cell) is not tested here; it runs on a networked box."""
|
||
|
|
import datetime
|
||
|
|
import math
|
||
|
|
|
||
|
|
import polars as pl
|
||
|
|
import pytest
|
||
|
|
|
||
|
|
import drift_check
|
||
|
|
from data import climate
|
||
|
|
|
||
|
|
_D = [datetime.date(2024, 1, d) for d in (1, 2, 3)]
|
||
|
|
|
||
|
|
|
||
|
|
def test_compare_values_stats():
|
||
|
|
a = pl.DataFrame({"date": _D, "tmax": [70.0, 80.0, 90.0], "wind": [5.0, 6.0, 7.0]})
|
||
|
|
b = pl.DataFrame({"date": _D, "tmax": [72.0, 78.0, 90.0], "wind": [5.0, 6.0, 7.0]})
|
||
|
|
out = drift_check.compare_values(a, b, metrics=("tmax", "wind"))
|
||
|
|
assert out["tmax"] == {"n": 3, "mad": pytest.approx(4 / 3, abs=1e-3),
|
||
|
|
"bias": pytest.approx(0.0), "max_abs": pytest.approx(2.0)}
|
||
|
|
assert out["wind"] == {"n": 3, "mad": 0.0, "bias": 0.0, "max_abs": 0.0}
|
||
|
|
|
||
|
|
|
||
|
|
def test_compare_values_skips_nulls_and_missing_metric():
|
||
|
|
a = pl.DataFrame({"date": _D, "tmax": [70.0, 80.0, 90.0]})
|
||
|
|
b = pl.DataFrame({"date": _D, "tmax": [71.0, None, 90.0]})
|
||
|
|
out = drift_check.compare_values(a, b, metrics=("tmax", "gust"))
|
||
|
|
assert out["tmax"]["n"] == 2 # the null day is dropped
|
||
|
|
assert out["tmax"]["mad"] == pytest.approx(0.5) # |70-71| and |90-90|
|
||
|
|
assert "gust" not in out # metric absent from both frames
|
||
|
|
|
||
|
|
|
||
|
|
def _gradeable(n=800):
|
||
|
|
"""A finalized, gradeable multi-year frame (seasonal temp swing so windows have a
|
||
|
|
real distribution)."""
|
||
|
|
start = datetime.date(2019, 1, 1)
|
||
|
|
dates = [start + datetime.timedelta(days=i) for i in range(n)]
|
||
|
|
tmax = [70.0 + 15.0 * math.sin(i / 58.0) for i in range(n)]
|
||
|
|
tmin = [t - 15.0 for t in tmax]
|
||
|
|
return climate._finalize_frame(pl.DataFrame({
|
||
|
|
"date": dates, "tmax": tmax, "tmin": tmin, "precip": [0.0] * n,
|
||
|
|
"wind": [5.0] * n, "gust": [8.0] * n, "humid": [60.0] * n,
|
||
|
|
"fmax": tmax, "fmin": tmin,
|
||
|
|
}))
|
||
|
|
|
||
|
|
|
||
|
|
def test_grade_band_divergence_identical_is_zero():
|
||
|
|
f = _gradeable()
|
||
|
|
end = f["date"].max()
|
||
|
|
start = end - datetime.timedelta(days=60)
|
||
|
|
div = drift_check.grade_band_divergence(f, f, start, end)
|
||
|
|
assert div # some metrics were compared
|
||
|
|
assert any(s["compared"] > 0 for s in div.values())
|
||
|
|
assert all(s["differ"] == 0 for s in div.values()) # identical frames never diverge
|
||
|
|
assert all(s["pct"] == 0.0 for s in div.values())
|