Merge pull request matplotlib#2996 from solvents/master

Violin Plots

Author: tacaswell

CHANGELOG

@@ -30,6 +30,10 @@
            interpolation = 'none' and interpolation = 'nearest' in
            `imshow()` when saving vector graphics files.
 
+2014-04-22 Added violin plotting functions. See `Axes.violinplot`,
+           `Axes.violin`, `cbook.violin_stats` and `mlab.GaussianKDE` for
+           details.
+
 2014-04-10 Fixed the triangular marker rendering error. The "Up" triangle was
            rendered instead of "Right" triangle and vice-versa.
 

boilerplate.py

@@ -146,6 +146,7 @@ def boilerplate_gen():
         'tricontourf',
         'tripcolor',
         'triplot',
+        'violinplot',
         'vlines',
         'xcorr',
         'barbs',

doc/users/whats_new.rst

@@ -172,6 +172,27 @@ Added the Axes method :meth:`~matplotlib.axes.Axes.add_image` to put image
 handling on a par with artists, collections, containers, lines, patches,
 and tables.
 
+Violin Plots
+````````````
+Per Parker, Gregory Kelsie, Adam Ortiz, Kevin Chan, Geoffrey Lee, Deokjae
+Donald Seo, and Taesu Terry Lim added a basic implementation for violin
+plots. Violin plots can be used to represent the distribution of sample data.
+They are similar to box plots, but use a kernel density estimation function to
+present a smooth approximation of the data sample used. The added features are:
+
+:func:`~matplotlib.Axes.violin` - Renders a violin plot from a collection of
+statistics.
+:func:`~matplotlib.cbook.violin_stats` - Produces a collection of statistics
+suitable for rendering a violin plot.
+:func:`~matplotlib.pyplot.violinplot` - Creates a violin plot from a set of
+sample data. This method makes use of :func:`~matplotlib.cbook.violin_stats`
+to process the input data, and :func:`~matplotlib.cbook.violin_stats` to
+do the actual rendering. Users are also free to modify or replace the output of
+:func:`~matplotlib.cbook.violin_stats` in order to customize the violin plots
+to their liking.
+
+This feature was implemented for a software engineering course at the
+University of Toronto, Scarborough, run in Winter 2014 by Anya Tafliovich.
 
 More `markevery` options to show only a subset of markers
 `````````````````````````````````````````````````````````

examples/statistics/violinplot_demo.py

@@ -0,0 +1,48 @@
+"""
+Demo of the new violinplot functionality
+"""
+
+import random
+import numpy as np
+import matplotlib.pyplot as plt
+
+# fake data
+fs = 10 # fontsize
+pos = [1,2,4,5,7,8]
+data = [np.random.normal(size=100) for i in pos]
+
+fig, axes = plt.subplots(nrows=2, ncols=3, figsize=(6,6))
+
+axes[0, 0].violinplot(data, pos, points=20, widths=0.1, 
+                      showmeans=True, showextrema=True, showmedians=True)
+axes[0, 0].set_title('Custom violinplot 1', fontsize=fs)
+
+axes[0, 1].violinplot(data, pos, points=40, widths=0.3, 
+                      showmeans=True, showextrema=True, showmedians=True,
+                      bw_method='silverman')
+axes[0, 1].set_title('Custom violinplot 2', fontsize=fs)
+
+axes[0, 2].violinplot(data, pos, points=60, widths=0.5, showmeans=True,
+                      showextrema=True, showmedians=True, bw_method=0.5)
+axes[0, 2].set_title('Custom violinplot 3', fontsize=fs)
+
+axes[1, 0].violinplot(data, pos, points=80, vert=False, widths=0.7,
+                      showmeans=True, showextrema=True, showmedians=True)
+axes[1, 0].set_title('Custom violinplot 4', fontsize=fs)
+
+axes[1, 1].violinplot(data, pos, points=100, vert=False, widths=0.9,
+                      showmeans=True, showextrema=True, showmedians=True,
+                      bw_method='silverman')
+axes[1, 1].set_title('Custom violinplot 5', fontsize=fs)
+
+axes[1, 2].violinplot(data, pos, points=200, vert=False, widths=1.1,
+                      showmeans=True, showextrema=True, showmedians=True,
+                      bw_method=0.5)
+axes[1, 2].set_title('Custom violinplot 6', fontsize=fs)
+
+for ax in axes.flatten():
+    ax.set_yticklabels([])
+
+fig.suptitle("Violin Plotting Examples")
+fig.subplots_adjust(hspace=0.4)
+plt.show()

lib/matplotlib/axes/_axes.py

@@ -6725,6 +6725,248 @@ def matshow(self, Z, **kwargs):
                                                  integer=True))
         return im
 
+    def violinplot(self, dataset, positions=None, vert=True, widths=0.5,
+                   showmeans=False, showextrema=True, showmedians=False,
+                   points=100, bw_method=None):
+        """
+        Make a violin plot.
+
+        Call signature::
+
+          violinplot(dataset, positions=None, vert=True, widths=0.5,
+                     showmeans=False, showextrema=True, showmedians=False,
+                     points=100, bw_method=None):
+
+        Make a violin plot for each column of *dataset* or each vector in
+        sequence *dataset*.  Each filled area extends to represent the
+        entire data range, with optional lines at the mean, the median,
+        the minimum, and the maximum.
+
+        Parameters
+        ----------
+
+          dataset : Array or a sequence of vectors.
+            The input data.
+
+          positions : array-like, default = [1, 2, ..., n]
+            Sets the positions of the violins. The ticks and limits are
+            automatically set to match the positions.
+
+          vert : bool, default = True.
+            If true, creates a vertical violin plot.
+            Otherwise, creates a horizontal violin plot.
+
+          widths : array-like, default = 0.5
+            Either a scalar or a vector that sets the maximal width of
+            each violin. The default is 0.5, which uses about half of the
+            available horizontal space.
+
+          showmeans : bool, default = False
+            If true, will toggle rendering of the means.
+
+          showextrema : bool, default = True
+            If true, will toggle rendering of the extrema.
+
+          showmedians : bool, default = False
+            If true, will toggle rendering of the medians.
+
+          points : scalar, default = 100
+            Defines the number of points to evaluate each of the gaussian
+            kernel density estimations at.
+
+          bw_method : str, scalar or callable, optional
+            The method used to calculate the estimator bandwidth.  This can be
+            'scott', 'silverman', a scalar constant or a callable.  If a
+            scalar, this will be used directly as `kde.factor`.  If a
+            callable, it should take a `GaussianKDE` instance as its only
+            parameter and return a scalar. If None (default), 'scott' is used.
+
+        Returns
+        -------
+
+        A dictionary mapping each component of the violinplot to a list of the
+        corresponding collection instances created. The dictionary has
+        the following keys:
+
+            - bodies: A list of the
+              :class:`matplotlib.collections.PolyCollection` instances
+              containing the filled area of each violin.
+            - means: A :class:`matplotlib.collections.LineCollection` instance
+              created to identify the mean values of each of the violin's
+              distribution.
+            - mins: A :class:`matplotlib.collections.LineCollection` instance
+              created to identify the bottom of each violin's distribution.
+            - maxes: A :class:`matplotlib.collections.LineCollection` instance
+              created to identify the top of each violin's distribution.
+            - bars: A :class:`matplotlib.collections.LineCollection` instance
+              created to identify the centers of each violin's distribution.
+            - medians: A :class:`matplotlib.collections.LineCollection`
+              instance created to identify the median values of each of the
+              violin's distribution.
+
+        """
+
+        def _kde_method(X, coords):
+            kde = mlab.GaussianKDE(X, bw_method)
+            return kde.evaluate(coords)
+
+        vpstats = cbook.violin_stats(dataset, _kde_method, points=points)
+        return self.violin(vpstats, positions=positions, vert=vert,
+                           widths=widths, showmeans=showmeans,
+                           showextrema=showextrema, showmedians=showmedians)
+
+    def violin(self, vpstats, positions=None, vert=True, widths=0.5,
+               showmeans=False, showextrema=True, showmedians=False):
+        """
+        Drawing function for violin plots.
+
+        Call signature::
+
+          violin(vpstats, positions=None, vert=True, widths=0.5,
+                 showmeans=False, showextrema=True, showmedians=False):
+
+        Draw a violin plot for each column of `vpstats`. Each filled area
+        extends to represent the entire data range, with optional lines at the
+        mean, the median, the minimum, and the maximum.
+
+        Parameters
+        ----------
+
+          vpstats : list of dicts
+            A list of dictionaries containing stats for each violin plot.
+            Required keys are:
+              - coords: A list of scalars containing the coordinates that
+                the violin's kernel density estimate were evaluated at.
+              - vals: A list of scalars containing the values of the kernel
+                density estimate at each of the coordinates given in `coords`.
+              - mean: The mean value for this violin's dataset.
+              - median: The median value for this violin's dataset.
+              - min: The minimum value for this violin's dataset.
+              - max: The maximum value for this violin's dataset.
+
+          positions : array-like, default = [1, 2, ..., n]
+            Sets the positions of the violins. The ticks and limits are
+            automatically set to match the positions.
+
+          vert : bool, default = True.
+            If true, plots the violins veritcally.
+            Otherwise, plots the violins horizontally.
+
+          widths : array-like, default = 0.5
+            Either a scalar or a vector that sets the maximal width of
+            each violin. The default is 0.5, which uses about half of the
+            available horizontal space.
+
+          showmeans : bool, default = False
+            If true, will toggle rendering of the means.
+
+          showextrema : bool, default = True
+            If true, will toggle rendering of the extrema.
+
+          showmedians : bool, default = False
+            If true, will toggle rendering of the medians.
+
+        Returns
+        -------
+
+        A dictionary mapping each component of the violinplot to a list of the
+        corresponding collection instances created. The dictionary has
+        the following keys:
+
+            - bodies: A list of the
+              :class:`matplotlib.collections.PolyCollection` instances
+              containing the filled area of each violin.
+            - means: A :class:`matplotlib.collections.LineCollection` instance
+              created to identify the mean values of each of the violin's
+              distribution.
+            - mins: A :class:`matplotlib.collections.LineCollection` instance
+              created to identify the bottom of each violin's distribution.
+            - maxes: A :class:`matplotlib.collections.LineCollection` instance
+              created to identify the top of each violin's distribution.
+            - bars: A :class:`matplotlib.collections.LineCollection` instance
+              created to identify the centers of each violin's distribution.
+            - medians: A :class:`matplotlib.collections.LineCollection`
+              instance created to identify the median values of each of the
+              violin's distribution.
+
+        """
+
+        # Statistical quantities to be plotted on the violins
+        means = []
+        mins = []
+        maxes = []
+        medians = []
+
+        # Collections to be returned
+        artists = {}
+
+        N = len(vpstats)
+        datashape_message = ("List of violinplot statistics and `{0}` "
+                             "values must have the same length")
+
+        # Validate positions
+        if positions is None:
+            positions = range(1, N + 1)
+        elif len(positions) != N:
+            raise ValueError(datashape_message.format("positions"))
+
+        # Validate widths
+        if np.isscalar(widths):
+            widths = [widths] * N
+        elif len(widths) != N:
+            raise ValueError(datashape_message.format("widths"))
+
+        # Calculate ranges for statistics lines
+        pmins = -0.25 * np.array(widths) + positions
+        pmaxes = 0.25 * np.array(widths) + positions
+
+        # Check whether we are rendering vertically or horizontally
+        if vert:
+            fill = self.fill_betweenx
+            perp_lines = self.hlines
+            par_lines = self.vlines
+        else:
+            fill = self.fill_between
+            perp_lines = self.vlines
+            par_lines = self.hlines
+
+        # Render violins
+        bodies = []
+        for stats, pos, width in zip(vpstats, positions, widths):
+            # The 0.5 factor reflects the fact that we plot from v-p to
+            # v+p
+            vals = np.array(stats['vals'])
+            vals = 0.5 * width * vals / vals.max()
+            bodies += [fill(stats['coords'],
+                            -vals + pos,
+                            vals + pos,
+                            facecolor='y',
+                            alpha=0.3)]
+            means.append(stats['mean'])
+            mins.append(stats['min'])
+            maxes.append(stats['max'])
+            medians.append(stats['median'])
+        artists['bodies'] = bodies
+
+        # Render means
+        if showmeans:
+            artists['cmeans'] = perp_lines(means, pmins, pmaxes, colors='r')
+
+        # Render extrema
+        if showextrema:
+            artists['cmaxes'] = perp_lines(maxes, pmins, pmaxes, colors='r')
+            artists['cmins'] = perp_lines(mins, pmins, pmaxes, colors='r')
+            artists['cbars'] = par_lines(positions, mins, maxes, colors='r')
+
+        # Render medians
+        if showmedians:
+            artists['cmedians'] = perp_lines(medians,
+                                             pmins,
+                                             pmaxes,
+                                             colors='r')
+
+        return artists
+
     def tricontour(self, *args, **kwargs):
         return mtri.tricontour(self, *args, **kwargs)
     tricontour.__doc__ = mtri.TriContourSet.tricontour_doc