add observed_rug argument to plot_ppc (#2161)

* add observed_rug argument to plot_ppc

* add tests and update changelog

CHANGELOG.md

@@ -4,8 +4,9 @@
 ## v0.x.x Unreleased
 
 ### New features
-* Add `weight_predictions` function to allow generation of weighted predictions from two or more InfereceData with `posterior_predictive` groups and a set of weights ([2147](https://github.com/arviz-devs/arviz/pull/2147))
+- Add `weight_predictions` function to allow generation of weighted predictions from two or more InfereceData with `posterior_predictive` groups and a set of weights ([2147](https://github.com/arviz-devs/arviz/pull/2147))
 - Add Savage-Dickey density ratio plot for Bayes factor approximation. ([2037](https://github.com/arviz-devs/arviz/pull/2037), [2152](https://github.com/arviz-devs/arviz/pull/2152
+- Adds rug plot for observed variables to `plot_ppc`. ([2161](https://github.com/arviz-devs/arviz/pull/2161))
 
 ### Maintenance and fixes
 - Fix dimension ordering for `plot_trace` with divergences ([2151](https://github.com/arviz-devs/arviz/pull/2151))

arviz/plots/backends/bokeh/ppcplot.py

@@ -1,11 +1,15 @@
 """Bokeh Posterior predictive plot."""
 import numpy as np
 from bokeh.models.annotations import Legend
+from bokeh.models.glyphs import Scatter
+from bokeh.models import ColumnDataSource
+
 
 from ....stats.density_utils import get_bins, histogram, kde
 from ...kdeplot import plot_kde
 from ...plot_utils import _scale_fig_size, vectorized_to_hex
 
+
 from .. import show_layout
 from . import backend_kwarg_defaults, create_axes_grid
 
@@ -27,6 +31,7 @@ def plot_ppc(
     textsize,
     mean,
     observed,
+    observed_rug,
     jitter,
     total_pp_samples,
     legend,  # pylint: disable=unused-argument
@@ -97,6 +102,7 @@ def plot_ppc(
         obs_vals = obs_vals.flatten()
         pp_vals = pp_vals.reshape(total_pp_samples, -1)
         pp_sampled_vals = pp_vals[pp_sample_ix]
+        cds_rug = ColumnDataSource({"_": np.array(obs_vals)})
 
         if kind == "kde":
             plot_kwargs = {
@@ -144,6 +150,16 @@ def plot_ppc(
                         return_glyph=True,
                     )
                     legend_it.append((label, glyph))
+                    if observed_rug:
+                        glyph = Scatter(
+                            x="_",
+                            y=0.0,
+                            marker="dash",
+                            angle=np.pi / 2,
+                            line_color=colors[1],
+                            line_width=linewidth,
+                        )
+                        ax_i.add_glyph(cds_rug, glyph)
                 else:
                     bins = get_bins(obs_vals)
                     _, hist, bin_edges = histogram(obs_vals, bins=bins)
@@ -215,6 +231,18 @@ def plot_ppc(
                         mode="center",
                     )
                     legend_it.append((label, [step]))
+
+                if observed_rug:
+                    glyph = Scatter(
+                        x="_",
+                        y=0.0,
+                        marker="dash",
+                        angle=np.pi / 2,
+                        line_color=colors[1],
+                        line_width=linewidth,
+                    )
+                    ax_i.add_glyph(cds_rug, glyph)
+
             pp_densities = np.empty((2 * len(pp_sampled_vals), pp_sampled_vals[0].size))
             for idx, vals in enumerate(pp_sampled_vals):
                 vals = np.array([vals]).flatten()

arviz/plots/backends/matplotlib/ppcplot.py

@@ -31,6 +31,7 @@ def plot_ppc(
     textsize,
     mean,
     observed,
+    observed_rug,
     jitter,
     total_pp_samples,
     legend,
@@ -135,6 +136,7 @@ def plot_ppc(
                 if dtype == "f":
                     plot_kde(
                         obs_vals,
+                        rug=observed_rug,
                         label="Observed",
                         plot_kwargs={"color": colors[1], "linewidth": linewidth, "zorder": 3},
                         fill_kwargs={"alpha": 0},
@@ -232,6 +234,14 @@ def plot_ppc(
                     drawstyle=drawstyle,
                     zorder=3,
                 )
+                if observed_rug:
+                    ax_i.plot(
+                        obs_vals,
+                        np.zeros_like(obs_vals) - 0.1,
+                        ls="",
+                        marker="|",
+                        color=colors[1],
+                    )
             if animated:
                 animate, init = _set_animation(
                     pp_sampled_vals,

arviz/plots/ppcplot.py

@@ -20,6 +20,7 @@ def plot_ppc(
     alpha=None,
     mean=True,
     observed=True,
+    observed_rug=False,
     color=None,
     colors=None,
     grid=None,
@@ -62,6 +63,9 @@ def plot_ppc(
         Defaults to ``True``.
     observed: bool, default True
         Whether or not to plot the observed data.
+    observed: bool, default False
+        Whether or not to plot a rug plot for the observed data. Only valid if `observed` is
+        `True` and for kind `kde` or `cumulative`.
     color: str
         Valid matplotlib ``color``. Defaults to ``C0``.
     color: list
@@ -339,6 +343,7 @@ def plot_ppc(
         textsize=textsize,
         mean=mean,
         observed=observed,
+        observed_rug=observed_rug,
         total_pp_samples=total_pp_samples,
         legend=legend,
         labeller=labeller,

arviz/tests/base_tests/test_plots_bokeh.py

@@ -878,12 +878,14 @@ def test_plot_violin_discrete(discrete_model):
 @pytest.mark.parametrize("kind", ["kde", "cumulative", "scatter"])
 @pytest.mark.parametrize("alpha", [None, 0.2, 1])
 @pytest.mark.parametrize("observed", [True, False])
-def test_plot_ppc(models, kind, alpha, observed):
+@pytest.mark.parametrize("observed_rug", [False, True])
+def test_plot_ppc(models, kind, alpha, observed, observed_rug):
     axes = plot_ppc(
         models.model_1,
         kind=kind,
         alpha=alpha,
         observed=observed,
+        observed_rug=observed_rug,
         random_seed=3,
         backend="bokeh",
         show=False,

arviz/tests/base_tests/test_plots_matplotlib.py

@@ -711,7 +711,8 @@ def test_plot_pair_shared(sharex, sharey, marginals):
 @pytest.mark.parametrize("alpha", [None, 0.2, 1])
 @pytest.mark.parametrize("animated", [False, True])
 @pytest.mark.parametrize("observed", [True, False])
-def test_plot_ppc(models, kind, alpha, animated, observed):
+@pytest.mark.parametrize("observed_rug", [False, True])
+def test_plot_ppc(models, kind, alpha, animated, observed, observed_rug):
     if animation and not animation.writers.is_available("ffmpeg"):
         pytest.skip("matplotlib animations within ArviZ require ffmpeg")
     animation_kwargs = {"blit": False}
@@ -720,6 +721,7 @@ def test_plot_ppc(models, kind, alpha, animated, observed):
         kind=kind,
         alpha=alpha,
         observed=observed,
+        observed_rug=observed_rug,
         animated=animated,
         animation_kwargs=animation_kwargs,
         random_seed=3,