Use contourpy for Bokeh contour plots

CHANGELOG.md

@@ -9,6 +9,8 @@
 * Add exception in `az.plot_hdi` for `x` of type `np.datetime64` and `smooth=True` ([2016](https://github.com/arviz-devs/arviz/pull/2016))
 * Change `ax.plot` usage to `ax.scatter` in `plot_pair` ([1990](https://github.com/arviz-devs/arviz/pull/1990))
 * Example data has been moved to the [arviz_example_data](https://github.com/arviz-devs/arviz_example_data) repository and is now included as a git submodule. ([2096](https://github.com/arviz-devs/arviz/pull/2096))
+* Bokeh kde contour plots started to use `contourpy` package ([2104](https://github.com/arviz-devs/arviz/pull/2104))
+* Update default Bokeh markers for rcparams ([2104](https://github.com/arviz-devs/arviz/pull/2104))
 
 ### Deprecation
 * Removed `fill_last`, `contour` and `plot_kwargs` arguments from `plot_pair` function ([2085](https://github.com/arviz-devs/arviz/pull/2085))

arviz/plots/backends/bokeh/kdeplot.py

@@ -3,7 +3,6 @@
 from collections.abc import Callable
 from numbers import Integral
 
-from matplotlib import _contour
 import numpy as np
 from bokeh.models import ColumnDataSource
 from bokeh.models.glyphs import Scatter
@@ -151,6 +150,13 @@ def plot_kde(
             glyphs.append(line)
 
     else:
+        try:
+            import contourpy
+        except ImportError as err:
+            raise ImportError(
+                "'bokeh' kde contour plots needs ContourPy installed (pip install countourpy)."
+            ) from err
+
         contour_kwargs = _init_kwargs_dict(contour_kwargs)
         contourf_kwargs = _init_kwargs_dict(contourf_kwargs)
         pcolormesh_kwargs = _init_kwargs_dict(pcolormesh_kwargs)
@@ -162,8 +168,10 @@ def plot_kde(
 
             scaled_density, *scaled_density_args = _scale_axis(density)
 
-            contour_generator = _contour.QuadContourGenerator(
-                x_x, y_y, scaled_density, None, True, 0
+            contourpy_kwargs = _init_kwargs_dict(contour_kwargs.pop("contourpy_kwargs", {}))
+            contourpy_kwargs.setdefault("name", "serial")
+            contour_generator = contourpy.contour_generator(
+                x=x_x, y=y_y, z=scaled_density, **contourpy_kwargs
             )
 
             levels = 9
@@ -199,7 +207,7 @@ def plot_kde(
                 contour_kwargs_ = contour_kwargs.copy()
                 contour_kwargs_.setdefault("line_color", color)
                 contour_kwargs_.setdefault("fill_color", color)
-                vertices, _ = contour_generator.create_filled_contour(level, level_upper)
+                vertices, _ = contour_generator.filled(level, level_upper)
                 for seg in vertices:
                     # ax.multi_polygon would be better, but input is
                     # currently not suitable

arviz/rcparams.py

@@ -153,23 +153,40 @@ def validate_with_none(value):
 
 def _validate_bokeh_marker(value):
     """Validate the markers."""
-    all_markers = (
-        "Asterisk",
-        "Circle",
-        "CircleCross",
-        "CircleX",
-        "Cross",
-        "Dash",
-        "Diamond",
-        "DiamondCross",
-        "Hex",
-        "InvertedTriangle",
-        "Square",
-        "SquareCross",
-        "SquareX",
-        "Triangle",
-        "X",
-    )
+    try:
+        from bokeh.core.enums import MarkerType
+    except ImportError:
+        MarkerType = [
+            "asterisk",
+            "circle",
+            "circle_cross",
+            "circle_dot",
+            "circle_x",
+            "circle_y",
+            "cross",
+            "dash",
+            "diamond",
+            "diamond_cross",
+            "diamond_dot",
+            "dot",
+            "hex",
+            "hex_dot",
+            "inverted_triangle",
+            "plus",
+            "square",
+            "square_cross",
+            "square_dot",
+            "square_pin",
+            "square_x",
+            "star",
+            "star_dot",
+            "triangle",
+            "triangle_dot",
+            "triangle_pin",
+            "x",
+            "y",
+        ]
+    all_markers = list(MarkerType)
     if value not in all_markers:
         raise ValueError(f"{value} is not one of {all_markers}")
     return value
@@ -274,7 +291,7 @@ def validate_iterable(value):
         "above",
         _make_validate_choice({"above", "below", "left", "right"}, allow_none=True),
     ),
-    "plot.bokeh.marker": ("Cross", _validate_bokeh_marker),
+    "plot.bokeh.marker": ("cross", _validate_bokeh_marker),
     "plot.bokeh.output_backend": ("webgl", _make_validate_choice({"canvas", "svg", "webgl"})),
     "plot.bokeh.show": (True, _validate_boolean),
     "plot.bokeh.tools": (

arviz/stats/stats.py

@@ -303,7 +303,7 @@ def gradient(weights):
             d_std_err = np.sqrt(len(diff) * np.var(diff))
             std_err = ses.loc[val]
             weight = weights[idx]
-            df_comp.at[val] = (
+            df_comp.loc[val] = (
                 idx,
                 res[f"elpd_{ic}"],
                 res[p_ic],
@@ -1941,7 +1941,7 @@ def apply_test_function(
         >>> import arviz as az
         >>> idata = az.load_arviz_data("centered_eight")
         >>> az.apply_test_function(idata, lambda y, theta: np.min(y))
-        >>> T = np.asscalar(idata.observed_data.T)
+        >>> T = idata.observed_data.T.item()
         >>> az.plot_posterior(idata, var_names=["T"], group="posterior_predictive", ref_val=T)
 
     """

arvizrc.template

@@ -37,7 +37,7 @@ plot.bokeh.layout.sizing_mode: fixed       # Responsive layout. One of "fixed",
                                            # "scale_width", "scale_height", "scale_both"
 plot.bokeh.layout.toolbar_location : above # Location for toolbar on layouts. "None" will hide the toolbar.
                                            # One of "above", "below", "left", "right".
-plot.bokeh.marker            : Cross       # specify the marker type used to plot
+plot.bokeh.marker            : cross       # specify the marker type used to plot
 plot.bokeh.output_backend    : webgl       # one of "canvas", "svg", "webgl"
 plot.bokeh.show              : true        # call bokeh.plotting.show for figure or grid of figures. One of "true", "false"
 plot.bokeh.tools             : reset,pan,box_zoom,wheel_zoom,lasso_select,undo,save,hover

requirements-docs.txt

@@ -11,6 +11,7 @@ sphinx-panels
 sphinx-notfound-page
 sphinx-copybutton
 bokeh
+contourpy
 sphinx_design
 sphinx-codeautolink>=0.9.0
 jupyter-sphinx

requirements-optional.txt

@@ -1,5 +1,6 @@
 numba
 bokeh>=1.4.0,<3.0
+contourpy
 ujson
 dask[distributed]
 zarr>=2.5.0