BUG: don't set aspect when plotting missing values (#3438)

Co-authored-by: Matt Richards <mrichards7@outlook.com.au>
geopandas · Nov 17, 2024 · e908d32 · e908d32
1 parent 2276f62
commit e908d32
Showing 3 changed files with 27 additions and 21 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -2,32 +2,16 @@
 
 ## Version 1.1.0
 
-Notes on dependencies:
-
-- GeoPandas 1.1 now requires Python 3.10 or greater and pandas 2.0, numpy 1.24, pyproj 3.5,
-  are now the minimum required version for these dependencies.
-  Furthermore, the minimum tested version for optional dependencies has been updated to
-  fiona 1.8.21, scipy 1.9, matplotlib 3.7, mapclassify 2.5, folium 0.12 and
-  SQLAlchemy 2.0. Older versions of these libraries may continue to work, but are no longer
-  considered supported (#3371).
-
 New features and improvements:
 
 - Add ``grid_size`` parameter to ``union_all`` and ``dissolve`` (#3445).
-
-Bug fixes:
-
-- Fix an issue that showed numpy dtypes in bbox in `to_geo_dict` and `__geo_interface__`. (#3436)
-- Fix an issue in `sample_points` that could occasionally result in non-uniform distribution (#3470)
-
-New features and improvements:
-
 - `GeoDataFrame.plot` now supports `pd.Index` as an input for the `column` keyword (#3463).
-
-## Version 1.0.2 (???)
+- Avoid change of the plot aspect when plotting missing values (#3438).
 
 Bug fixes:
 
+- Fix an issue that showed numpy dtypes in bbox in `to_geo_dict` and `__geo_interface__`. (#3436)
+- Fix an issue in `sample_points` that could occasionally result in non-uniform distribution (#3470).
 - Fix unspecified layer warning being emitted while reading multilayer datasets, even
   when layer is specified when using the mask or bbox keywords (#3378).
 - Properly support named aggregations over a geometry column in `GroupBy.agg` (#3368).
@@ -38,7 +22,18 @@ Bug fixes:
 - Fix regression where `overlay` with `keep_geom_type` returns wrong results if the
   input contains invalid geometries (#3395).
 - Fix the dtype of the GeometryArray backing data being incorrect for zero length
-  GeoDataFrames causing errors in `overlay` (3424).
+  GeoDataFrames causing errors in `overlay` (#3424).
+- Fix regression where constructing a GeoSeries from a pd.Series with GeometryDtype values
+  failed when `crs` was provided (#3383).
+
+Notes on dependencies:
+
+- GeoPandas 1.1 now requires Python 3.10 or greater and pandas 2.0, numpy 1.24, pyproj 3.5,
+  are now the minimum required version for these dependencies.
+  Furthermore, the minimum tested version for optional dependencies has been updated to
+  fiona 1.8.21, scipy 1.9, matplotlib 3.7, mapclassify 2.5, folium 0.12 and
+  SQLAlchemy 2.0. Older versions of these libraries may continue to work, but are no longer
+  considered supported (#3371).
 
 Deprecations and compatibility notes:
 

diff --git a/geopandas/plotting.py b/geopandas/plotting.py
@@ -882,7 +882,7 @@ def plot_dataframe(
         merged_kwds = style_kwds.copy()
         merged_kwds.update(missing_kwds)
 
-        plot_series(expl_series[nan_idx], ax=ax, **merged_kwds)
+        plot_series(expl_series[nan_idx], ax=ax, **merged_kwds, aspect=None)
 
     if legend and not color:
         if legend_kwds is None:

diff --git a/geopandas/tests/test_plotting.py b/geopandas/tests/test_plotting.py
@@ -451,6 +451,17 @@ def test_no_missing_and_missing_kwds(self):
         df["category"] = df["values"].astype("str")
         df.plot("category", missing_kwds={"facecolor": "none"}, legend=True)
 
+    def test_missing_aspect(self):
+        self.df.loc[0, "values"] = np.nan
+        ax = self.df.plot(
+            "values",
+            missing_kwds={"color": "r"},
+            categorical=True,
+            legend=True,
+            aspect=2,
+        )
+        assert ax.get_aspect() == 2
+
 
 class TestPointZPlotting:
     def setup_method(self):