Add box-plots to standard deviation plot

- test for stddev plot
- Update baseline image for test_that_all_snake_oil_visualisations_matches_snapshot
- Remove redudant image files

Co-authored by Feda Curic <[email protected]>

src/ert/gui/plottery/plots/std_dev.py

@@ -1,4 +1,4 @@
-from typing import Any, Dict, List
+from typing import Any, Dict
 
 import matplotlib.pyplot as plt
 import numpy as np
@@ -28,40 +28,88 @@ def plot(
         if layer is not None:
             vmin: float = np.inf
             vmax: float = -np.inf
-            axes = []
-            images: List[npt.NDArray[np.float32]] = []
+            heatmaps = []
+            boxplot_axes = []
+
+            # Adjust height_ratios to reduce space between plots
+            figure.set_layout_engine("constrained")
+            gridspec = figure.add_gridspec(2, ensemble_count, hspace=0.2)
+
             for i, ensemble in enumerate(plot_context.ensembles(), start=1):
-                ax = figure.add_subplot(1, ensemble_count, i)
-                axes.append(ax)
+                ax_heat = figure.add_subplot(gridspec[0, i - 1])
+                ax_box = figure.add_subplot(gridspec[1, i - 1])
                 data = std_dev_data[ensemble.name]
                 if data.size == 0:
-                    ax.set_axis_off()
-                    ax.text(
+                    ax_heat.set_axis_off()
+                    ax_box.set_axis_off()
+                    ax_heat.text(
                         0.5,
                         0.5,
                         f"No data for {ensemble.experiment_name} : {ensemble.name}",
                         ha="center",
                         va="center",
                     )
                 else:
-                    images.append(data)
                     vmin = min(vmin, float(np.min(data)))
                     vmax = max(vmax, float(np.max(data)))
-                ax.set_title(
+
+                    im = ax_heat.imshow(data, cmap="viridis", aspect="equal")
+                    heatmaps.append(im)
+
+                    ax_box.boxplot(data.flatten(), vert=True, widths=0.5)
+                    boxplot_axes.append(ax_box)
+
+                    min_value = np.min(data)
+                    mean_value = np.mean(data)
+                    max_value = np.max(data)
+
+                    annotation_text = f"Min: {min_value:.2f}\nMean: {mean_value:.2f}\nMax: {max_value:.2f}"
+                    ax_box.annotate(
+                        annotation_text,
+                        xy=(1, 1),  # Changed from (0, 1) to (1, 1)
+                        xycoords="axes fraction",
+                        ha="right",  # Changed from 'left' to 'right'
+                        va="top",
+                        fontsize=8,
+                        fontweight="bold",
+                        bbox={
+                            "facecolor": "white",
+                            "edgecolor": "black",
+                            "boxstyle": "round,pad=0.2",
+                        },
+                    )
+
+                    ax_box.spines["top"].set_visible(False)
+                    ax_box.spines["right"].set_visible(False)
+                    ax_box.spines["bottom"].set_visible(False)
+                    ax_box.spines["left"].set_visible(True)
+
+                    ax_box.set_xticks([])
+                    ax_box.set_xticklabels([])
+
+                    ax_heat.set_ylabel("")
+                    ax_box.set_ylabel(
+                        "Standard Deviation", fontsize=8
+                    )  # Reduced font size
+
+                    self._colorbar(im)
+
+                ax_heat.set_title(
                     f"{ensemble.experiment_name} : {ensemble.name} layer={layer}",
                     wrap=True,
+                    fontsize=10,  # Reduced font size
                 )
 
             norm = plt.Normalize(vmin, vmax)
-            for ax, data in zip(axes, images):
-                if data is not None:
-                    im = ax.imshow(data, norm=norm, cmap="viridis")
-                    self._colorbar(im)
-            figure.tight_layout()
+            for im in heatmaps:
+                im.set_norm(norm)
+
+            padding = 0.05 * (vmax - vmin)
+            for ax_box in boxplot_axes:
+                ax_box.set_ylim(vmin - padding, vmax + padding)
 
     @staticmethod
     def _colorbar(mappable: Any) -> Any:
-        # https://joseph-long.com/writing/colorbars/
         last_axes = plt.gca()
         ax = mappable.axes
         assert ax is not None

tests/ert/ui_tests/gui/test_plotting_of_snake_oil.py

@@ -23,14 +23,14 @@
 from ert.services import StorageService
 from ert.storage import open_storage
 
-from .conftest import (
-    get_child,
-    wait_for_child,
-)
+from .conftest import get_child, wait_for_child
 
 
 # Use a fixture for the figure in order for the lifetime
-# of the c++ gui element to not go out before mpl_image_compare
+# of the c++ gui element to not go out before mpl_image_compare.
+# Note that the data is copied from test-data and all the existing storages
+# there will be copied too! They need to be removed!
+# Once the storage is created it its cached in .pytest_cache.
 @pytest.fixture(
     params=[
         ("FOPR", STATISTICS, "snake_oil"),
@@ -97,6 +97,12 @@ def plot_figure(qtbot, heat_equation_storage, snake_oil_case_storage, request):
                                 selected_key.dimensionality
                                 == tab._plotter.dimensionality
                             )
+                            if plot_name == STD_DEV:
+                                # we need a better resolution for box plots
+                                tab._figure.set_size_inches(
+                                    2000 / tab._figure.get_dpi(),
+                                    1000 / tab._figure.get_dpi(),
+                                )
                             yield tab._figure.figure
                         else:
                             assert (

tests/ert/unit_tests/gui/plottery/test_stddev_plot.py

@@ -0,0 +1,49 @@
+from unittest.mock import Mock
+
+import matplotlib.pyplot as plt
+import numpy as np
+import pytest
+from matplotlib.figure import Figure
+
+from ert.gui.plottery import PlotConfig, PlotContext
+from ert.gui.plottery.plots.std_dev import StdDevPlot
+from ert.gui.tools.plot.plot_api import EnsembleObject
+
+
+@pytest.fixture()
+def plot_context(request):
+    context = Mock(spec=PlotContext)
+    context.ensembles.return_value = [
+        EnsembleObject("ensemble_1", "id", False, "experiment_1")
+    ]
+    context.history_data = None
+    context.layer = 0
+    context.plotConfig.return_value = PlotConfig(title="StdDev Plot")
+    return context
+
+
+def test_stddev_plot_shows_boxplot(plot_context: PlotContext):
+    rng = np.random.default_rng()
+    figure = Figure()
+    std_dev_data = rng.random((5, 5))
+    StdDevPlot().plot(
+        figure,
+        plot_context,
+        {},
+        {},
+        {"ensemble_1": std_dev_data},
+    )
+    ax = figure.axes
+    assert ax[0].get_title() == "experiment_1 : ensemble_1 layer=0"
+    assert ax[1].get_ylabel() == "Standard Deviation"
+    annotation = [
+        child for child in ax[1].get_children() if isinstance(child, plt.Annotation)
+    ]
+    assert len(annotation) == 1
+    min_value = np.min(std_dev_data)
+    mean_value = np.mean(std_dev_data)
+    max_value = np.max(std_dev_data)
+    assert (
+        annotation[0].get_text()
+        == f"Min: {min_value:.2f}\nMean: {mean_value:.2f}\nMax: {max_value:.2f}"
+    )