Plot a histogram with constant bar widths but different bin sizes

Question

I have a Python script which produces a single histogram of three data sets, where the bin sizes are provided as a list. The Python script sippet is as follows:

bin_l=np.arange(0,14,1)
bin_l=np.append(bin_l,max_bin)

plt.hist([wind_plot_o, wind_plot_m, wind_plot_b],bins=bin_l, align='mid',
     label=['label1', 'label2', 'label3'],color=['k','g', 'r'])

Where, the element appended to bin_l later is the maximum of the data given. So the last bin size would be different than other bin sizes.
The above script snippet, along with the rest of the script, produces the following plot:
Which is good, except for a problem. I need all the bars to be of same size, I do not want the bar sizes to depend upon the bin size, which doesn't look good in my case.
The matplotlib documentation for hist() says the parameter 'rwidth' can be used to adjust the bar width, still relative to the bin size, which is what it is currently doing and of no use to me.
So, how can I keep the bar widths of matplotlib histogram same while using non-uniform bin sizes?

Note: It may look like a duplicate of this post, but it isn't. I am not using logarithmic scale for the X axis.

I would not recommend using unequal bin sizes when creating grouped histograms. This makes the histogram very hard or even impossible to read. Why not use bin_l=np.arange(0,17,1) from the start? — ImportanceOfBeingErnest
@ImportanceOfBeingErnest The part in the unequal bins are not that important, but I wouldn't want them to be excluded. That's why I am using an unequal bin size at the end. — RogUE

busybear busybear · Accepted Answer · 2019-02-04T18:45:58

This is a bit unconventional to do. But plt.hist returns the number of elements in each bin. You can use this data (or from numpy.histogram to skip plotting the original histogram) to create a "bar graph" with equal widths instead.

counts, _, _ = plt.hist(...)
plt.close()                                # Don't want the plot from plt.hist
for n, Y in enumerate(counts):
    X = [n*.3 + x for x in range(len(Y))]  # position bars next to each other
    plt.bar(X, Y, 0.3)

Plot a histogram with constant bar widths but different bin sizes

1 Answers