Fix detect_outliers_ecod to handle pandas Series and missing values

src/growthcurves/preprocessing.py

@@ -251,6 +251,12 @@ def detect_outliers_ecod(N: np.ndarray, factor: float = 3.5) -> np.ndarray:
         Boolean mask of the same length as N where True indicates an outlier.
     """
 
+    N = np.asarray(N, dtype=float)
+    if np.any(np.isnan(N)):
+        raise ValueError(
+            "detect_outliers_ecod does not support missing values (NaN). "
+            "Remove or impute NaN values before calling this function."
+        )
     n = len(N)
     half = 15 // 2
     residual = np.zeros(n)

tests/test_preprocessing.py

@@ -0,0 +1,53 @@
+import numpy as np
+import pandas as pd
+import pytest
+
+from growthcurves.preprocessing import detect_outliers_ecod
+
+
+def _sample_series():
+    """Return a simple OD-like growth series."""
+    return [0.1, 0.12, 0.15, 0.2, 0.3, 0.5, 0.8, 1.2, 1.5, 1.6, 1.61, 1.62]
+
+
+def test_ecod_pandas_series_default_index():
+    """detect_outliers_ecod should accept a pandas Series with the default (0-based) index."""
+    s = pd.Series(_sample_series())
+    result = detect_outliers_ecod(s)
+    assert isinstance(result, np.ndarray)
+    assert result.dtype == bool
+    assert len(result) == len(s)
+
+
+def test_ecod_pandas_series_non_default_index():
+    """detect_outliers_ecod should accept a pandas Series with a non-standard index."""
+    values = _sample_series()
+    # Use an index that starts at 5 so label-based access N[0] would raise KeyError
+    s = pd.Series(values, index=range(5, 5 + len(values)))
+    result = detect_outliers_ecod(s)
+    assert isinstance(result, np.ndarray)
+    assert result.dtype == bool
+    assert len(result) == len(s)
+
+
+def test_ecod_numpy_array():
+    """detect_outliers_ecod should work with a plain numpy array."""
+    arr = np.array(_sample_series())
+    result = detect_outliers_ecod(arr)
+    assert isinstance(result, np.ndarray)
+    assert result.dtype == bool
+    assert len(result) == len(arr)
+
+
+def test_ecod_missing_values_raises():
+    """detect_outliers_ecod should raise ValueError when the input contains NaN."""
+    arr = np.array([0.1, 0.2, np.nan, 0.4, 0.5] * 3, dtype=float)
+    with pytest.raises(ValueError, match="missing values"):
+        detect_outliers_ecod(arr)
+
+
+def test_ecod_missing_values_in_series_raises():
+    """detect_outliers_ecod should raise ValueError when a pandas Series contains NaN."""
+    s = pd.Series([0.1, 0.2, np.nan, 0.4, 0.5] * 3)
+    with pytest.raises(ValueError, match="missing values"):
+        detect_outliers_ecod(s)