New to MatPlotLib! How to round the edges of the bars in a Bar-Plot in MatPlotLib?

Community
1

I’m using MatPlotLib to do a Bar-Plot. The only thing that I want to add, is rounded edges to the bars. From what I understand there isn’t a direct way of doing it, in MatPlotLib. Anyone has any idea how i can do it?

Here is my code:

(...)
#extract the values for selected columns
        player_values = player[columns_to_plot]

        #create a dark grey figure
        fig, ax = plt.subplots(figsize=(24, 16), constrained_layout=True)
        fig.set_facecolor('#333333')

        #set the colors for each range
        colors = ['#FF6666', '#FFA366', '#FFFF99', '#99FF99']
        color_thresholds = [1, 25, 50, 75]

        #adjust the width of the bars and separation
        bar_width = 0.1
        bar_separation = 0.05

        #add labels on top of each bar with color coding
        for i, (col, value) in enumerate(zip(columns_to_plot_labels, player_values)):
            color_index = np.digitize(value, color_thresholds)-1
            rect = ax.bar(i*(bar_width+bar_separation), value, color=colors[color_index],
                          width=bar_width, capstyle='round')
            if not np.isnan(value):
                ax.text(i*(bar_width+bar_separation), value+0.5, str(int(value)),
                        ha='center', va='bottom', color='white', weight='bold', fontsize=28)
(...)

Thanks!

2

There don’t appear to be any good ways. Depending on the range of your data, you may find these two options usable:

A) a fancy box, which is what I’ve done below. It has some problems with short bars - try removing the if statement.
B) a slightly more complex thing where you reduce the height of each bar, add a circle patch at each corner, then add an additional rectangle if needed to fill any gaps. Bit of a faff either way.

full code:

import numpy as np
from matplotlib.patches import FancyBboxPatch, Rectangle

columns_to_plot_labels = ['a', 'b', 'c', 'd', 'e', 'f']
player_values = [20, 5, 80, 30, None, 55]

# Handle null values _before_ plotting your data
values_replacing_nans = [x if x is not None else 0 for x in player_values ]
max_col_height = max(values_replacing_nans)
min_col_height = min(values_replacing_nans)

# create a dark grey figure
fig, ax = plt.subplots(figsize=(6,4), constrained_layout=True)
fig.set_facecolor('#333333')
ax.set_facecolor('#333333')

# set the colors for each range
colors = ['#FF6666', '#FFA366', '#FFFF99', '#99FF99'] 
color_thresholds = [1, 25, 50, 75]

# Set the rounding values
bar_width = 0.5
bar_separation = 0.1
bar_pad_factor = 0  # proportion of bar_width to add for rounding, 0 means no additional width
bar_rounding_factor = 0.2 # proportion of bar_width to add for rounding, 0 means no rounding, 0.5 means a circlular top

# Plot all bars together
bars = ax.bar(
    x=columns_to_plot_labels, 
    height=values_replacing_nans, 
    color=[colors[c] for c in np.digitize(values_replacing_nans, color_thresholds)-1],
    )

# loop through the bars
for i, bar in enumerate(bars):
    # add the text on top of the bars
    ax.text(bar.get_x() + bar.get_width()/2, bar.get_height()+0.5, values_replacing_nans[i],
        ha='center', va='bottom', color='white', weight='bold', fontsize=28)
    
    if bar.get_height() > bar_rounding_factor*max_col_height:
        # add the rounded bar on top of the original bar, note - this is rounded at both ends
        round_top = FancyBboxPatch(
            xy = bar.get_xy(),  # use the original bar's values to fill the new bar
            width=bar.get_width(), 
            height=bar.get_height(), 
            color=bar.get_facecolor(),
            boxstyle=f"round,pad={bar_pad_factor},rounding_size={bar_rounding_factor}",
            transform=ax.transData,
            mutation_scale=1.1,
            mutation_aspect=20,
            )
        # Over write the bottom half of the original bar with a Rectangle patch
        square_bottom = Rectangle(
            xy = bar.get_xy(), 
            width=bar.get_width(), 
            height=bar.get_height()/2, 
            color=bar.get_facecolor(),
            transform=ax.transData,
            )
        # remove the original bar from the plot
        bar.remove()
        # add the new artists to the plot
        ax.add_patch(round_top)
        ax.add_patch(square_bottom)
plt.show()

Which produces this plot:
image

1 Like