BUG: Fixes #36918 boxplots with matplotlib

doc/source/whatsnew/v1.2.0.rst

@@ -437,6 +437,7 @@ Plotting
 - Bug in :meth:`DataFrame.plot` was rotating xticklabels when ``subplots=True``, even if the x-axis wasn't an irregular time series (:issue:`29460`)
 - Bug in :meth:`DataFrame.plot` where a marker letter in the ``style`` keyword sometimes causes a ``ValueError`` (:issue:`21003`)
 - Twinned axes were losing their tick labels which should only happen to all but the last row or column of 'externally' shared axes (:issue:`33819`)
+- Bug in :meth:`DataFrame.boxplot` was raising ``ValueError`` when plotting with ``vert=True`` on a subplot with shared axes (:issue:`36918`)
 
 Groupby/resample/rolling
 ^^^^^^^^^^^^^^^^^^^^^^^^

pandas/plotting/_matplotlib/boxplot.py

@@ -246,6 +246,7 @@ def boxplot(
 ):
 
     import matplotlib.pyplot as plt
+    from matplotlib.ticker import FixedFormatter
 
     # validate return_type:
     if return_type not in BoxPlot._valid_return_types:
@@ -302,15 +303,13 @@ def plot_group(keys, values, ax: "Axes"):
         bp = ax.boxplot(values, **kwds)
         if fontsize is not None:
             ax.tick_params(axis="both", labelsize=fontsize)
+
         if kwds.get("vert", 1):
-            ticks = ax.get_xticks()
-            if len(ticks) != len(keys):
-                i, remainder = divmod(len(ticks), len(keys))
-                assert remainder == 0, remainder
-                keys *= i
-            ax.set_xticklabels(keys, rotation=rot)
+            axis = ax.xaxis
         else:
-            ax.set_yticklabels(keys, rotation=rot)
+            axis = ax.yaxis
+        axis.set_major_formatter(FixedFormatter(keys))
+        ax.tick_params(axis=axis.axis_name, which="major", rotation=rot)
         maybe_color_bp(bp, **kwds)
 
         # Return axes in multiplot case, maybe revisit later # 985

pandas/tests/plotting/test_boxplot_method.py

@@ -218,6 +218,40 @@ def test_specified_props_kwd(self, props, expected):
 
         assert result[expected][0].get_color() == "C1"
 
+    @pytest.mark.parametrize("vert", [(True), (False)])
+    def test_boxplot_shared_axis(self, vert):
+        # GH 37107
+        df1 = DataFrame(np.random.random((100, 5)), columns=["A", "B", "C", "D", "E"])
+        df2 = DataFrame(np.random.random((100, 5)), columns=["A", "B", "C", "D", "E"])
+
+        # Two rows if shared axis is y, two rows if shared axis is y.
+        # This is done so that the shared axes are actually separated
+        # and get_ticklabels returns a non empty list on each ax object
+        nrows, ncols, sharex, sharey = (
+            (1, 2, True, False) if vert else (2, 1, False, True)
+        )
+        fig, axes = self.plt.subplots(
+            nrows=nrows, ncols=ncols, sharex=sharex, sharey=sharey
+        )
+        df1.boxplot(ax=axes[0], vert=vert, fontsize=10, rot=10)
+        df2.boxplot(ax=axes[1], vert=vert, fontsize=10, rot=10)
+
+        # In order for the ticklabels to be placed, the plot has to be drawn
+        fig.canvas.draw()
+
+        for ax in axes:
+            axis = ax.xaxis if vert else ax.yaxis
+            labels = [x.get_text() for x in axis.get_ticklabels(which="major")]
+            if vert:
+                self._check_ticks_props(ax, xlabelsize=10, xrot=10)
+            else:
+                self._check_ticks_props(ax, ylabelsize=10, yrot=10)
+
+            # Matplotlib returns 10 ticklabels, 5 of which are empty
+            assert len(labels) % 5 == 0
+            assert len(labels) // (len(labels) // 5) == 5
+            assert labels[:5] == ["A", "B", "C", "D", "E"]
+
 
 @td.skip_if_no_mpl
 class TestDataFrameGroupByPlots(TestPlotBase):