matplotlib · jklymak · Jul 20, 2021 · Jul 14, 2021
diff --git a/examples/statistics/barchart_demo.py b/examples/statistics/barchart_demo.py
@@ -15,23 +15,16 @@
 just make up some data for little Johnny Doe.
 """

+from collections import namedtuple
 import numpy as np
 import matplotlib.pyplot as plt
-from matplotlib.ticker import MaxNLocator
-from collections import namedtuple

-np.random.seed(42)

 Student = namedtuple('Student', ['name', 'grade', 'gender'])
-Score = namedtuple('Score', ['score', 'percentile'])
+Score = namedtuple('Score', ['value', 'unit', 'percentile'])

-# GLOBAL CONSTANTS
-test_names = ['Pacer Test', 'Flexed Arm\n Hang', 'Mile Run', 'Agility',
-              'Push Ups']
-test_units = dict(zip(test_names, ['laps', 'sec', 'min:sec', 'sec', '']))

-
-def attach_ordinal(num):
+def to_ordinal(num):
    """Convert an integer to an ordinal string, e.g. 2 -> '2nd'."""
    suffixes = {str(i): v
                for i, v in enumerate(['th', 'st', 'nd', 'rd', 'th',
@@ -43,118 +36,68 @@ def attach_ordinal(num):
    return v + suffixes[v[-1]]


-def format_score(score, test):
+def format_score(score):
    """
    Create score labels for the right y-axis as the test name followed by the
    measurement unit (if any), split over two lines.
    """
-    unit = test_units[test]
-    if unit:
-        return f'{score}\n{unit}'
-    else:  # If no unit, don't include a newline, so that label stays centered.
-        return score
-
+    return f'{score.value}\n{score.unit}' if score.unit else str(score.value)

-def format_ycursor(y):
-    y = int(y)
-    if y < 0 or y >= len(test_names):
-        return ''
-    else:
-        return test_names[y]

-
-def plot_student_results(student, scores, cohort_size):
-    fig, ax1 = plt.subplots(figsize=(9, 7))  # Create the figure
-    fig.subplots_adjust(left=0.115, right=0.88)
+def plot_student_results(student, scores_by_test, cohort_size):
+    fig, ax1 = plt.subplots(figsize=(9, 7), constrained_layout=True)
    fig.canvas.manager.set_window_title('Eldorado K-8 Fitness Chart')

-    pos = np.arange(len(test_names))
-
-    rects = ax1.barh(pos, [scores[k].percentile for k in test_names],
-                     align='center',
-                     height=0.5,
-                     tick_label=test_names)
-
    ax1.set_title(student.name)
+    ax1.set_xlabel(
+        'Percentile Ranking Across {grade} Grade {gender}s\n'
+        'Cohort Size: {cohort_size}'.format(
+            grade=to_ordinal(student.grade),
+            gender=student.gender.title(),
+            cohort_size=cohort_size))
+
+    test_names = list(scores_by_test.keys())
+    percentiles = [score.percentile for score in scores_by_test.values()]
+
+    rects = ax1.barh(test_names, percentiles, align='center', height=0.5)
+    # Partition the percentile values to be able to draw large numbers in
+    # white within the bar, and small numbers in black outside the bar.
+    large_percentiles = [to_ordinal(p) if p > 40 else '' for p in percentiles]
+    small_percentiles = [to_ordinal(p) if p <= 40 else '' for p in percentiles]
+    ax1.bar_label(rects, small_percentiles,
+                  padding=5, color='black', fontweight='bold')
+    ax1.bar_label(rects, large_percentiles,
+                  padding=-32, color='white', fontweight='bold')

    ax1.set_xlim([0, 100])
-    ax1.xaxis.set_major_locator(MaxNLocator(11))
+    ax1.set_xticks([0, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100])
    ax1.xaxis.grid(True, linestyle='--', which='major',
                   color='grey', alpha=.25)
-
-    # Plot a solid vertical gridline to highlight the median position
-    ax1.axvline(50, color='grey', alpha=0.25)
+    ax1.axvline(50, color='grey', alpha=0.25)  # median position

    # Set the right-hand Y-axis ticks and labels
    ax2 = ax1.twinx()
-
-    # Set the tick locations and labels
-    ax2.set_yticks(
-        pos, labels=[format_score(scores[k].score, k) for k in test_names])
    # Set equal limits on both yaxis so that the ticks line up
    ax2.set_ylim(ax1.get_ylim())
+    # Set the tick locations and labels
+    ax2.set_yticks(
+        np.arange(len(scores_by_test)),
+        labels=[format_score(score) for score in scores_by_test.values()])

    ax2.set_ylabel('Test Scores')

-    xlabel = ('Percentile Ranking Across {grade} Grade {gender}s\n'
-              'Cohort Size: {cohort_size}')
-    ax1.set_xlabel(xlabel.format(grade=attach_ordinal(student.grade),
-                                 gender=student.gender.title(),
-                                 cohort_size=cohort_size))
-
-    rect_labels = []
-    # Lastly, write in the ranking inside each bar to aid in interpretation
-    for rect in rects:
-        # Rectangle widths are already integer-valued but are floating
-        # type, so it helps to remove the trailing decimal point and 0 by
-        # converting width to int type
-        width = int(rect.get_width())
-
-        rank_str = attach_ordinal(width)
-        # The bars aren't wide enough to print the ranking inside
-        if width < 40:
-            # Shift the text to the right side of the right edge
-            xloc = 5
-            # Black against white background
-            clr = 'black'
-            align = 'left'
-        else:
-            # Shift the text to the left side of the right edge
-            xloc = -5
-            # White on magenta
-            clr = 'white'
-            align = 'right'
-
-        # Center the text vertically in the bar
-        yloc = rect.get_y() + rect.get_height() / 2
-        label = ax1.annotate(
-            rank_str, xy=(width, yloc), xytext=(xloc, 0),
-            textcoords="offset points",
-            horizontalalignment=align, verticalalignment='center',
-            color=clr, weight='bold', clip_on=True)
-        rect_labels.append(label)
-
-    # Make the interactive mouse over give the bar title
-    ax2.fmt_ydata = format_ycursor
-    # Return all of the artists created
-    return {'fig': fig,
-            'ax': ax1,
-            'ax_right': ax2,
-            'bars': rects,
-            'perc_labels': rect_labels}
-
-
-student = Student('Johnny Doe', 2, 'boy')
-scores = dict(zip(
-    test_names,
-    (Score(v, p) for v, p in
-     zip(['7', '48', '12:52', '17', '14'],
-         np.round(np.random.uniform(0, 100, len(test_names)), 0)))))
-cohort_size = 62  # The number of other 2nd grade boys
-
-arts = plot_student_results(student, scores, cohort_size)
-plt.show()

+student = Student(name='Johnny Doe', grade=2, gender='Boy')
+scores_by_test = {
+    'Pacer Test': Score(7, 'laps', percentile=37),
+    'Flexed Arm\n Hang': Score(48, 'sec', percentile=95),
+    'Mile Run': Score('12:52', 'min:sec', percentile=73),
+    'Agility': Score(17, 'sec', percentile=60),
+    'Push Ups': Score(14, '', percentile=16),
+}
+
+plot_student_results(student, scores_by_test, cohort_size=62)
+plt.show()

 #############################################################################
 #
@@ -164,5 +107,5 @@ def plot_student_results(student, scores, cohort_size):
 #    in this example:
 #
 #    - `matplotlib.axes.Axes.bar` / `matplotlib.pyplot.bar`
-#    - `matplotlib.axes.Axes.annotate` / `matplotlib.pyplot.annotate`
+#    - `matplotlib.axes.Axes.bar_label` / `matplotlib.pyplot.bar_label`
 #    - `matplotlib.axes.Axes.twinx` / `matplotlib.pyplot.twinx`