Pandas: can't plot multi-row subplots

Created on 26 Aug 2014 · 3Comments · Source: pandas-dev/pandas

This code works fine to produce a single row of 3 subplots

g = np.random.choice([1,2,3], 10)
s = np.random.normal(size=10)
s2 =np.random.normal(size=10)
df = pd.DataFrame([g, s, s2]).T
df.columns = ['key', 's1', 's2']
gb = df.groupby('key')

fig, axes = plt.subplots(nrows=1, ncols=3, figsize=(12, 4))
i = 0
for key, df2 in gb:
    df2.plot(ax=axes[i], x='s1', y='s2', title=key)
    i = i + 1

But if I try to add a second row (nrows=2), it blows up with an Attribute Error

AttributeError: 'numpy.ndarray' object has no attribute 'get_figure'

fig, axes = plt.subplots(nrows=2, ncols=3)
fig.tight_layout() # Or equivalently,  "plt.tight_layout()"
i = 1
for key, df2 in gb:
    df2.plot(ax=axes[i])
    i = i + 1

Source

ericdf

👍1

Most helpful comment

If you want to wrap around to the second row you'll have to do something like axes[i // 3][i % 3]
(I expanded your example to have 6 groups)

In [67]: df
Out[67]: 
    key        s1        s2
0     3 -1.452043 -0.119374
1     1  0.603860 -1.635034
2     3  0.964165 -0.043124
3     2  0.459628 -0.538155
4     3  0.398761 -0.195261
5     1  0.085750 -0.116766
6     2 -0.397419 -0.140660
7     3 -0.053209  1.547755
8     1 -0.634555 -0.509077
9     3  0.138808  0.608165
10    6 -1.452043 -0.119374
11    4  0.603860 -1.635034
12    6  0.964165 -0.043124
13    5  0.459628 -0.538155
14    6  0.398761 -0.195261
15    4  0.085750 -0.116766
16    5 -0.397419 -0.140660
17    6 -0.053209  1.547755
18    4 -0.634555 -0.509077
19    6  0.138808  0.608165

In [63]: for i, (key, df2) in enumerate(gb):
    df2.plot(ax=axes[i // 3][i % 3])

TomAugspurger on 26 Aug 2014

👍4 ❤2

All 3 comments

That's because axes is now a 2-d array of matplotlib axes.

In [35]: fig, axes = plt.subplots(nrows=2, ncols=3)

In [36]: axes
Out[36]: 
array([[<matplotlib.axes._subplots.AxesSubplot object at 0x10c094ac8>,
        <matplotlib.axes._subplots.AxesSubplot object at 0x10ee40b70>,
        <matplotlib.axes._subplots.AxesSubplot object at 0x10c0ac240>],
       [<matplotlib.axes._subplots.AxesSubplot object at 0x10edf8a90>,
        <matplotlib.axes._subplots.AxesSubplot object at 0x10ec27630>,
        <matplotlib.axes._subplots.AxesSubplot object at 0x10edc9128>]], dtype=object)

In [37]: axes.shape
Out[37]: (2, 3)

Try something like

In [38]: for i, (key, df2) in enumerate(gb):
    df2.plot(ax=axes[0][i])

TomAugspurger on 26 Aug 2014

👍2

If you want to wrap around to the second row you'll have to do something like axes[i // 3][i % 3]
(I expanded your example to have 6 groups)

In [67]: df
Out[67]: 
    key        s1        s2
0     3 -1.452043 -0.119374
1     1  0.603860 -1.635034
2     3  0.964165 -0.043124
3     2  0.459628 -0.538155
4     3  0.398761 -0.195261
5     1  0.085750 -0.116766
6     2 -0.397419 -0.140660
7     3 -0.053209  1.547755
8     1 -0.634555 -0.509077
9     3  0.138808  0.608165
10    6 -1.452043 -0.119374
11    4  0.603860 -1.635034
12    6  0.964165 -0.043124
13    5  0.459628 -0.538155
14    6  0.398761 -0.195261
15    4  0.085750 -0.116766
16    5 -0.397419 -0.140660
17    6 -0.053209  1.547755
18    4 -0.634555 -0.509077
19    6  0.138808  0.608165

In [63]: for i, (key, df2) in enumerate(gb):
    df2.plot(ax=axes[i // 3][i % 3])

TomAugspurger on 26 Aug 2014

👍4 ❤2

Thank you so much, Tom. This now works perfectly. Sincerely appreciate your helpful reply.

ericdf on 27 Aug 2014

Was this page helpful?

0 / 5 - 0 ratings