sort bars in ascending order without involving dataFrames

Question

I have two lists and I am plotting a graph using matplotlib. Currently, the bars are organized in the way the lists are written. I want to set them in an ascending/descending order automatically. How can I do so?

industries = ['Manufacturing', 'Food', 'Eco']
counts = [12,78,1]

plt.figure(figsize=(16,6))
bars = plt.bar(industries, counts, width=0.2, bottom=None, align='center', data=None)
plt.xlim(-0.9, len(industries) - 1 + 0.9)
for i in range(len(counts)):
    percentage = ((counts[i]/(total))*100)
    plt.annotate(percentage, xy=(industries[i], counts[i] + 10), ha='center')
plt.show()

Edit:

I realized that the bars are built on the basis of alphabetical order. Even if the data is sorted. How can this be fixed?

Have you been able to resolve all your bar plotting questions now? — Trenton McKinney, Aug 03 '20 at 17:54

Dan · Answer 1 · 2020-08-02T22:56:50.427

0

First sort the lists as you need them by ziping them, sort, and then "unzip" them:

industries = ['Manufacturing', 'Food', 'Eco']
counts = [12,78,1]

sorted_counts, sorted_industries = zip(*sorted(zip(counts, industries), reverse=True))

Or set reversed=False for ascending.

Then plot like this

x_locations = range(len(industries))
plt.bar(x_locations, sorted_counts, , tick_label=sorted_industries, ...)

If you look at the docs you'll see the first parameter is not the bar labels but there x-coordinates. The simplest way to get them evenly spaced is to use range. Then you can change the tick lables using the tick_labels argument.

Just to note, when I try it using plt.bar(sorted_industries, sorted_counts), it is plotting them in the expected order, not alphabetical. What version of matplotlib are you trying this with?

edited Aug 02 '20 at 22:56

answered Aug 02 '20 at 22:41

Dan

45,079
17
88
157

Ah yes that makes sense too but sorting the list isn't the problem. Matplotlib automatically arranges everything in an alphabetical order even when the lists are sorted – Aug 02 '20 at 22:43
oh I see - yeah you should use `range(len(industries))` as the x-values, I'll update – Dan Aug 02 '20 at 22:45
@a125 see my edit on how to plot it in the correct order – Dan Aug 02 '20 at 22:52

Trenton McKinney · Accepted Answer · 2020-08-02T23:07:19.623

The two lists need to be combined with zip and then sorted on the counts.
Iterate through the list of tuples to add the bars, in sorted order in reverse order, with the annotation.

industries = ['Manufacturing', 'Food', 'Eco']
counts = [12, 78, 1]
tot = sum(counts)

# combine the two lists with zip and then reverse sort them
data = sorted(zip(industries, counts), key=lambda v: v[1], reverse=True)

plt.figure(figsize=(16, 6))
for (i, c) in data:  # unpack and plot each tuple in sorted order
    bars = plt.bar(i, c, width=0.2, bottom=None, align='center', data=None, color='g')
    plt.annotate(f'{(c/tot)*100:0.02f}%\n', xy=(i, c), va='center', ha='center')
plt.xlim(-0.9, len(industries) - 1 + 0.9)

plt.show()

sort bars in ascending order without involving dataFrames

2 Answers2