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

Question

Why is this matplotlib code giving me a weird exception? I'm going for two rows of plots. The top row is supposed to show true vs. pred and the bottom row is supposed to show percent error.

yy = func(*X)
fig, axes = plt.subplots(1, len(X))
for ax,_x in zip(axes,X):
 ax.plot(_x, y, 'b.')
 ax.plot(_x, yy, 'r.')
fig, axes = plt.subplots(2, len(X))
for ax,_x in zip(axes,X):
 ax.plot(_x, yy/y-1, 'r.')
plt.show()

Traceback:

 File "pysr.py", line 235, in main
 ax.plot(_x, yy/y-1, 'r.')
AttributeError: 'numpy.ndarray' object has no attribute 'plot'

• Blue_vision: Could you give a little more information on what X, y, and func are? I'm not able to reproduce the error you showed. Just a thought, if the code you are running has a typo and is instead for ax, _x in zip(X, axes), that would reproduce the error (and be a typo I could totally see myself making.)

Answer 1

If len(X) is>1, axes will be a 2D array of AxesSubplot instances. So when you loop over axes, you actually get a slice along one dimension of the axes array.

To overcome this, you could use axes.flat:

for ax,_x in zip(axes.flat,X):

Also if you are trying to plot all these on one figure, you don't need to call plt.subplots twice, as that will create two figures.

It may be easier to index the axes array like this:

yy = func(*X)
fig, axes = plt.subplots(2, len(X))
for i,_x in enumerate(X):
 axes[0, i].plot(_x, y, 'b.')
 axes[0, i].plot(_x, yy, 'r.')
 axes[1, i].plot(_x, yy/y-1, 'r.')
plt.show()