fix(datasets) Fix the scale of value axis when plotting in absolute s…

…izes (#4255)

datasets/flwr_datasets/visualization/comparison_label_distribution.py

@@ -15,7 +15,7 @@
 """Comparison of label distribution plotting."""
 
 
-from typing import Any, Optional, Union
+from typing import Any, Literal, Optional, Union
 
 import matplotlib.colors as mcolors
 import matplotlib.pyplot as plt
@@ -33,10 +33,10 @@
 def plot_comparison_label_distribution(
     partitioner_list: list[Partitioner],
     label_name: Union[str, list[str]],
-    plot_type: str = "bar",
-    size_unit: str = "percent",
-    max_num_partitions: Optional[Union[int]] = 30,
-    partition_id_axis: str = "y",
+    plot_type: Literal["bar", "heatmap"] = "bar",
+    size_unit: Literal["percent", "absolute"] = "percent",
+    max_num_partitions: Optional[int] = 30,
+    partition_id_axis: Literal["x", "y"] = "y",
     figsize: Optional[tuple[float, float]] = None,
     subtitle: str = "Comparison of Per Partition Label Distribution",
     titles: Optional[list[str]] = None,
@@ -55,14 +55,14 @@ def plot_comparison_label_distribution(
         List of partitioners to be compared.
     label_name : Union[str, List[str]]
         Column name or list of column names identifying labels for each partitioner.
-    plot_type : str
+    plot_type :  Literal["bar", "heatmap"]
         Type of plot, either "bar" or "heatmap".
-    size_unit : str
+    size_unit : Literal["percent", "absolute"]
         "absolute" for raw counts, or "percent" to normalize values to 100%.
     max_num_partitions : Optional[int]
         Maximum number of partitions to include in the plot. If None, all partitions
         are included.
-    partition_id_axis : str
+    partition_id_axis : Literal["x", "y"]
         Axis on which the partition IDs will be marked, either "x" or "y".
     figsize : Optional[Tuple[float, float]]
         Size of the figure. If None, a default size is calculated.
@@ -151,7 +151,10 @@ def plot_comparison_label_distribution(
             f"{type(label_name)}"
         )
     figsize = _initialize_comparison_figsize(figsize, num_partitioners)
-    fig, axes = plt.subplots(1, num_partitioners, layout="constrained", figsize=figsize)
+    axes_sharing = _initialize_axis_sharing(size_unit, plot_type, partition_id_axis)
+    fig, axes = plt.subplots(
+        1, num_partitioners, layout="constrained", figsize=figsize, **axes_sharing
+    )
 
     if titles is None:
         titles = ["" for _ in range(num_partitioners)]
@@ -201,11 +204,12 @@ def plot_comparison_label_distribution(
         axis.set_xlabel("")
         axis.set_ylabel("")
         axis.set_title(titles[idx])
-    for axis in axes[1:]:
-        axis.set_yticks([])
+    _set_tick_on_value_axes(axes, partition_id_axis, size_unit)
 
     # Set up figure xlabel and ylabel
-    xlabel, ylabel = _initialize_comparison_xy_labels(plot_type, partition_id_axis)
+    xlabel, ylabel = _initialize_comparison_xy_labels(
+        plot_type, size_unit, partition_id_axis
+    )
     fig.supxlabel(xlabel)
     fig.supylabel(ylabel)
     fig.suptitle(subtitle)
@@ -226,11 +230,13 @@ def _initialize_comparison_figsize(
 
 
 def _initialize_comparison_xy_labels(
-    plot_type: str, partition_id_axis: str
+    plot_type: Literal["bar", "heatmap"],
+    size_unit: Literal["percent", "absolute"],
+    partition_id_axis: Literal["x", "y"],
 ) -> tuple[str, str]:
     if plot_type == "bar":
         xlabel = "Partition ID"
-        ylabel = "Class distribution"
+        ylabel = "Class distribution" if size_unit == "percent" else "Class Count"
     elif plot_type == "heatmap":
         xlabel = "Partition ID"
         ylabel = "Label"
@@ -243,3 +249,34 @@ def _initialize_comparison_xy_labels(
         xlabel, ylabel = ylabel, xlabel
 
     return xlabel, ylabel
+
+
+def _initialize_axis_sharing(
+    size_unit: Literal["percent", "absolute"],
+    plot_type: Literal["bar", "heatmap"],
+    partition_id_axis: Literal["x", "y"],
+) -> dict[str, bool]:
+    # Do not intervene when the size_unit is percent and plot_type is heatmap
+    if size_unit == "percent":
+        return {}
+    if plot_type == "heatmap":
+        return {}
+    if partition_id_axis == "x":
+        return {"sharey": True}
+    if partition_id_axis == "y":
+        return {"sharex": True}
+    return {"sharex": False, "sharey": False}
+
+
+def _set_tick_on_value_axes(
+    axes: list[Axes],
+    partition_id_axis: Literal["x", "y"],
+    size_unit: Literal["percent", "absolute"],
+) -> None:
+    if partition_id_axis == "x" and size_unit == "absolute":
+        # Exclude this case due to sharing of y-axis (and thus y-ticks)
+        # They must remain set and the number are displayed only on the first plot
+        pass
+    else:
+        for axis in axes[1:]:
+            axis.set_yticks([])