Matplotlib: How to get space between bars?

Question

Hi I have started using matplotlib and have been trying to adapt the example code on the website to suit my needs. I have the code below which does what I want apart from the 3rd bar in each group overlaps the first of the next group of bars. Internet isnt good enough to add picture but any help would be great and if you could explain what my error is that would be appreciated.

Thanks, Tom

"""
Bar chart demo with pairs of bars grouped for easy comparison.
"""
import numpy as np
import matplotlib.pyplot as plt


n_groups = 3

means_e1 = (20, 35, 30)
std_e1 = (2, 3, 4)

means_e2 = (25, 32, 34)
std_e2 = (3, 5, 2)

means_e3 = (5, 2, 4)
std_e3 = (0.3, 0.5, 0.2)

fig, ax = plt.subplots()

index = np.arange(n_groups)
bar_width = 0.35

opacity = 0.4
error_config = {'ecolor': '0.3'}

rects1 = plt.bar(index , means_e1, bar_width,
                 alpha=opacity,
                 color='b',
                 yerr=std_e1,
                 error_kw=error_config,
                 label='Main')

rects2 = plt.bar(index + bar_width + 0.1, means_e2, bar_width,
                 alpha=opacity,
                 color='r',
                 yerr=std_e2,
                 error_kw=error_config,
                 label='e2')

rects3 = plt.bar(index + bar_width + bar_width + 0.2, means_e3, bar_width,
                 alpha=opacity,
                 color='g',
                 yerr=std_e3,
                 error_kw=error_config,
                 label='e3')

plt.xlabel('Dataset type used')
plt.ylabel('Percentage of reads joined after normalisation to 1 million reads')
plt.title('Application of Thimble on datasets, showing the ability of each stitcher option.')
plt.xticks(index + bar_width + bar_width, ('1', '2', '3'))
plt.legend()

plt.tight_layout()
plt.show()

Answer 1

The bar_width + bar_width + 0.2 is 0.9. Now you add another bar of bar_width (0.35), so overall you have 1.25, which is greater than 1. Since 1 is the distance between subsequent points of the index, you have overlap.

You can either increase the distance between index (index = np.arange(0, n_groups * 2, 2)), or reduce the bar width to something smaller, say 0.2.