add percentages to bar plots

src/skrubview/_plotting.py

@@ -6,20 +6,23 @@
 from . import _utils
 
 # from matplotlib import colormaps, colors
-# _PASTEL = list(map(colors.rgb2hex, colormaps.get_cmap("tab10").colors))
+# _TAB10 = list(map(colors.rgb2hex, colormaps.get_cmap("tab10").colors))
 
+
+# sns.color_palette('muted').as_hex()
 _SEABORN = [
-    "#4c72b0",
-    "#dd8452",
-    "#55a868",
-    "#c44e52",
-    "#8172b3",
-    "#937860",
-    "#da8bc3",
-    "#8c8c8c",
-    "#ccb974",
-    "#64b5cd",
+    "#4878d0",
+    "#ee854a",
+    "#6acc64",
+    "#d65f5f",
+    "#956cb4",
+    "#8c613c",
+    "#dc7ec0",
+    "#797979",
+    "#d5bb67",
+    "#82c6e2",
 ]
+
 COLORS = _SEABORN
 COLOR_0 = COLORS[0]
 
@@ -80,7 +83,7 @@ def line(x_col, y_col):
     return _serialize(fig)
 
 
-def value_counts(value_counts, n_unique, color=COLOR_0):
+def value_counts(value_counts, n_unique, n_rows, color=COLOR_0):
     values = [_utils.ellide_string_short(s) for s in value_counts.keys()][::-1]
     counts = list(value_counts.values())[::-1]
     if n_unique > len(value_counts):
@@ -89,7 +92,16 @@ def value_counts(value_counts, n_unique, color=COLOR_0):
         title = None
     fig, ax = plt.subplots()
     _despine(ax)
-    ax.barh(list(map(str, range(len(values)))), counts, color=color)
+    rects = ax.barh(list(map(str, range(len(values)))), counts, color=color)
+    percent = [_utils.format_percent(c / n_rows) for c in counts]
+    large_percent = [
+        f"{p: >6}" if c > counts[-1] / 2 else "" for (p, c) in zip(percent, counts)
+    ]
+    small_percent = [
+        p if c <= counts[-1] / 2 else "" for (p, c) in zip(percent, counts)
+    ]
+    ax.bar_label(rects, large_percent, padding=-30, color="black", fontsize=8)
+    ax.bar_label(rects, small_percent, padding=5, color="black", fontsize=8)
     ax.set_yticks(ax.get_yticks())
     ax.set_yticklabels(list(map(str, values)))
     if title is not None:

src/skrubview/_summarize.py

@@ -115,7 +115,7 @@ def _add_value_counts(summary, column, *, dataframe_summary, with_plots):
         summary["value_is_constant"] = False
         if with_plots:
             summary["value_counts_plot"] = _plotting.value_counts(
-                value_counts, n_unique, color=_plotting.COLORS[1]
+                value_counts, n_unique, dataframe_summary["n_rows"], color=_plotting.COLORS[1]
             )