86 lines
3.0 KiB
Python
86 lines
3.0 KiB
Python
"""Tests para describe_numeric."""
|
|
|
|
import os
|
|
import sys
|
|
|
|
sys.path.insert(0, os.path.dirname(__file__))
|
|
|
|
from describe_numeric import describe_numeric
|
|
|
|
# Keys that every result dict must always contain (the eda numeric_sub contract).
|
|
_EXPECTED_KEYS = {
|
|
"min", "max", "mean", "median", "mode", "std", "variance", "cv",
|
|
"p1", "p5", "p25", "p50", "p75", "p95", "p99", "iqr",
|
|
"skew", "kurtosis", "n_outliers", "outlier_pct",
|
|
"zero_pct", "negative_pct", "distribution_type", "histogram",
|
|
}
|
|
|
|
|
|
def test_lista_con_outlier_y_none():
|
|
"""Lista con outlier claro y None descartado."""
|
|
# Tight cluster around 2-4 plus a None to drop and a clear extreme outlier.
|
|
# A wide cluster (n=40) keeps std small so the extreme value's z-score
|
|
# exceeds the 3.0 threshold used by detect_outliers.
|
|
cluster = [1, 2, 2, 3, 4] * 8 # 40 numeric values, mode == 2
|
|
values = cluster + [None, 1000]
|
|
result = describe_numeric(values)
|
|
|
|
# Contract: all keys present.
|
|
assert set(result.keys()) == _EXPECTED_KEYS
|
|
|
|
# Non-numeric / missing dropped: 41 numeric values remain.
|
|
assert result["min"] == 1.0
|
|
assert result["max"] == 1000.0
|
|
|
|
# mean/median reasonable: median sits in the cluster, mean pulled up by 1000.
|
|
assert result["median"] < result["mean"]
|
|
assert 0.0 < result["median"] <= 5.0
|
|
assert result["mean"] > result["median"]
|
|
|
|
# mode = most frequent (2 appears twice per block).
|
|
assert result["mode"] == 2.0
|
|
|
|
# At least one z-score outlier detected (the 1000).
|
|
assert result["n_outliers"] >= 1
|
|
assert result["outlier_pct"] > 0.0
|
|
|
|
# Histogram non-empty and counts cover every numeric value.
|
|
assert len(result["histogram"]) > 0
|
|
total = sum(bucket["count"] for bucket in result["histogram"])
|
|
assert total == 41
|
|
for bucket in result["histogram"]:
|
|
assert "lo" in bucket and "hi" in bucket and "count" in bucket
|
|
|
|
# No zeros, no negatives in this sample.
|
|
assert result["zero_pct"] == 0.0
|
|
assert result["negative_pct"] == 0.0
|
|
|
|
|
|
def test_lista_vacia_todo_none():
|
|
"""Lista vacia (o sin numericos) devuelve todas las claves en None."""
|
|
result = describe_numeric([None, "abc", float("nan")])
|
|
|
|
assert set(result.keys()) == _EXPECTED_KEYS
|
|
for key in _EXPECTED_KEYS - {"histogram"}:
|
|
assert result[key] is None, f"{key} debe ser None"
|
|
assert result["histogram"] == []
|
|
|
|
|
|
def test_cv_none_cuando_mean_cero():
|
|
"""cv es None cuando la media es 0."""
|
|
# Symmetric around zero so mean == 0.
|
|
result = describe_numeric([-2, -1, 0, 1, 2])
|
|
assert result["mean"] == 0.0
|
|
assert result["cv"] is None
|
|
assert result["zero_pct"] == 20.0
|
|
assert result["negative_pct"] == 40.0
|
|
|
|
|
|
def test_iqr_y_percentiles():
|
|
"""iqr = p75 - p25 y percentiles coherentes."""
|
|
result = describe_numeric(list(range(1, 101))) # 1..100
|
|
assert result["iqr"] == result["p75"] - result["p25"]
|
|
assert result["p1"] <= result["p25"] <= result["p50"] <= result["p75"] <= result["p99"]
|
|
assert result["min"] == 1.0
|
|
assert result["max"] == 100.0
|