Python matplotlib stacked bar chart -- strange results

Question

I'm trying to plot four stacked bar charts on the same plot using Python's matplotlib library.

For each observation (obs1, obs2, obs3, obs4), I want to view the quantity of each component (c1, c2, c3, c4, c5, c6) using a stacked bar chart. This is the code I have written:

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

data = np.array([[-904., 97., 59., 5., 252., 138.], [-603., 65., 0., 29., 0., 0.], [-571., -27., 0., -28., 0., 0.], [-80., 40., 0., -9., 0., 0.]])

data2 = pd.DataFrame(data=data)
data2.index = ['obs1', 'obs2', 'obs3', 'obs4']
data2.columns = ['c1', 'c2', 'c3', 'c4', 'c5', 'c6']

ind = np.arange(4)
width = 0.4

p1 = plt.bar(ind, data2['c1'], width)
p2 = plt.bar(ind, data2['c2'], width)
p3 = plt.bar(ind, data2['c3'], width)
p4 = plt.bar(ind, data2['c4'], width)
p5 = plt.bar(ind, data2['c5'], width)
p6 = plt.bar(ind, data2['c6'], width)

plt.legend((p1[0], p2[0], p3[0], p4[0], p5[0], p6[0]), tuple(data2.columns), bbox_to_anchor = (1.05, 1), loc = 'upper left', borderaxespad = 0.)

For clarity, this is the DataFrame (used to produce the plot):

print(data2)
         c1    c2    c3    c4     c5     c6
obs1 -904.0  97.0  59.0   5.0  252.0  138.0
obs2 -603.0  65.0   0.0  29.0    0.0    0.0
obs3 -571.0 -27.0   0.0 -28.0    0.0    0.0
obs4  -80.0  40.0   0.0  -9.0    0.0    0.0

This is the plot:

Note on the plot, the bar for obs1 is at x=0, the bar for obs2 is at x=1, and so on.

However, there are two problems:

obs1 has a value of 252 for component 5, but the height of component 5 (in purple) is substantially below 252. How can I fix this?
obs3 has a value of -27 for component 2, but this is not shown on the plot at all. How can I solve this?

Thanks.

Toukenize · Accepted Answer · 2020-02-26 03:12:07Z

This is because when you plot, the y-value reference is taken from y=0, thus the purple bar for obs1 component 5 actually spanned from y=0 to y=252, and it got blocked by bar for component 6 (plotted later) due to the sequential nature matplotlib plots the patches of bars.

Similary, obs3 component 2 is not shown for the same reason.

To get what you want, a stacked bar chart, you can simply plot using pandas plotting interface:

fig, ax = plt.subplots(1,1,figsize=(6,4))

data2.plot(kind='bar', stacked=True, ax=ax)

ax.set_xticklabels(ax.get_xticklabels(), rotation=0)

plt.show()

which gives you exactly what you want:

On a side note, I would suggest to plot a separate bar chart for each component, as they are of very different scale, this would make the component level comparison across observations much clearer:

fig, ax = plt.subplots(1, 6, figsize=(20,4))

for index, col in enumerate(data2.columns):
    data2[col].plot(kind='bar', ax=ax[index], color=f'C{index}', title=f'{col}')
    ax[index].grid(True)
    ax[index].set_xticklabels(ax[index].get_xticklabels(), rotation=0)

plt.show()

which gives you:

Collectives™ on Stack Overflow

Python matplotlib stacked bar chart -- strange results

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related