Merge pull request #562 from DHI/test/no_index_name_change_in_input

Don't mutate inputs

Author: ecomodeller

src/modelskill/comparison/_comparison.py

@@ -341,10 +341,12 @@ def _matched_data_to_xarray(
                 f"Model data: {m} is of type {df[m].dtype}, it must be numeric"
             )
 
-    df = df[items.all]
-    df.index.name = "time"
-    df = df.rename(columns={items.obs: "Observation"})
-    ds = df.to_xarray()
+    ds = (
+        df.loc[:, items.all]
+        .rename_axis("time")
+        .rename(columns={items.obs: "Observation"})
+        .to_xarray()
+    )
     assert isinstance(ds, xr.Dataset)
 
     ds.attrs["name"] = name if name is not None else items.obs

tests/test_comparer.py

@@ -94,6 +94,8 @@ def tc() -> Comparer:
 
 
 def test_matched_df(pt_df):
+    # modelskill doesn't care about the name of the index, but it should be preserved
+    pt_df.index.name = "dato_tid"
     cmp = Comparer.from_matched_data(data=pt_df)
     assert cmp.gtype == "point"
     assert "m2" in cmp.mod_names
@@ -104,6 +106,14 @@ def test_matched_df(pt_df):
     assert cmp.score()["m1"] == pytest.approx(0.5916079783099617)
     assert cmp.score()["m2"] == pytest.approx(0.15811388300841905)
 
+    # from_matched doesn't modify the original dataframe,. including the name of the index
+    assert pt_df.index.name == "dato_tid"
+    assert list(pt_df.columns) == ["Observation", "m1", "m2"]
+
+    # but to_dataframe always returns a dataframe with index named "time"
+    df2 = cmp.to_dataframe()
+    assert df2.index.name == "time"
+
 
 def test_matched_skill_geodataframe(pt_df):
     cmp = Comparer.from_matched_data(data=pt_df, x=10.0, y=55.0)