Matplotlib bar chart with multiple columns and string X axis

Question

I have two parallel series of data, for which I want to make a bar chart with multiple columns. I understand the idea is to shift each column by adding the width the x axis, but what if the axis consists in strings (categorical variables)?

Let's assume this is a pandas dataframe df:

+----------+------+------+
| Category | Val1 | Val2 |
+----------+------+------+
| A        |    1 |    3 |
| B        |    4 |    6 |
| C        |    6 |    7 |
| D        |    3 |    9 |
+----------+------+------+

Obviously, this will raise an error:

import matplotlib.pyplot as plt
width = 0.4
fig, ax = plt.subplots()
ax.bar(x=df['Category'], height=df['Val1'], width=width)
ax.bar(x=df['Category']+width, height=df['Val2'], width=width)

You can't add a numeral to a string.

So, what's the solution to get something like this: bar chart

score 4 · Accepted Answer · answered May 15 '20 at 13:55

4

You could use pandas plot function:

fig, ax = plt.subplots()
df.plot.bar(x='Category', ax=ax)

Output:

answered May 15 '20 at 13:55

Quang Hoang

146,074
10
56
74

1

Ah, yes! Just for the sake of learning, if anyone can suggest a solution in pure matplotlib, I'd be very interested. – mrgou May 15 '20 at 14:11
1

For a pure matplotlib, shift the bars by `+/- width/2`. – Quang Hoang May 15 '20 at 14:12
1

I know that's the idea, but you can't with categorical variables. You can't calculate a shift based on a string. – mrgou May 15 '20 at 14:25
Just plot against `np.arange(len(df))` and `set_xticks`. – Quang Hoang May 15 '20 at 14:28

Matplotlib bar chart with multiple columns and string X axis

1 Answers1