Question

I am using a confusion matrix to measure the performance of my classifier. This example would work fine for me (its from here), but I get the whole time TypeError: Invalid dimensions for image data

from numpy import *
import matplotlib.pyplot as plt
from pylab import *

conf_arr = [[50.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0], [3.0, 26.0, 0.0, 0.0, 0.0, 1.0, 0.0], [0.0, 0.0, 10.0, 0.0, 0.0, 0.0, 0.0], [4.0, 1.0, 0.0, 5.0, 0.0, 0.0, 0.0], [3.0, 0.0, 1.0, 0.0, 6.0, 0.0, 0.0], [0.0, 0.0, 0.0, 0.0, 0.0, 47.0, 0.0], [2.0, 0.0, 0.0, 0.0, 0.0, 0.0, 8.0]]

norm_conf = []
for i in conf_arr:
        a = 0
        tmp_arr = []
        a = sum(i,0)
        for j in i:
                tmp_arr.append(float(j)/float(a))
        norm_conf.append(tmp_arr)

plt.clf()
fig = plt.figure()
ax = fig.add_subplot(111)
res = ax.imshow(array(norm_conf), cmap=cm.jet, interpolation='nearest')
cb = fig.colorbar(res)
savefig("confmat.png", format="png")

I am new to python and matplotlib. Any help?

Matplot version is 1.1.1. and here is the full traceback:

after res =… I get

TypeError      Traceback (most recent call last)
    C:\Python27\lib\site-packages\SimpleCV\Shell\Shell.pyc in <module>()
    ----> 1 res = ax.imshow(array(norm_conf), cmap=cm.jet, interpolation='nearest')

C:\Python27\lib\site-packages\matplotlib\axes.pyc in imshow(self, X, cmap, norm, aspect, interpolation, alpha, vmin, vmax, origin, extent, shape, filternorm, filter
rad, imlim, resample, url, **kwargs)
   6794                        filterrad=filterrad, resample=resample, **kwargs)
   6795
-> 6796         im.set_data(X)
   6797         im.set_alpha(alpha)
   6798         self._set_artist_props(im)

C:\Python27\lib\site-packages\matplotlib\image.pyc in set_data(self, A)
    409         if (self._A.ndim not in (2, 3) or
    410             (self._A.ndim == 3 and self._A.shape[-1] not in (3, 4))):
--> 411             raise TypeError("Invalid dimensions for image data")
    412
    413         self._imcache =None

TypeError: Invalid dimensions for image data

SimpleCV:105> cb = fig.colorbar(res)

For print norm_conf I get now results: [[1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0], [1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0], [1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],...]]. I corrected the indentation problem. But my confusion .png is pretty distorted. Further how should I proceed to label the squares in the matrix?

Answer 1

This works for me just fine (matplotlib 1.1.1rc). Originally I wanted you to confirm your matplotlib version and post the entire traceback — by “traceback” I mean the few lines before the TypeError line which show what caused the error — and that’s still a good idea, but I think I see what the problem might be.

This is the error you’d get if norm_conf somehow weren’t being filled (i.e. norm_conf = []):

Traceback (most recent call last):
  File "mdim2.py", line 19, in <module>
    res = ax.imshow(array(norm_conf), cmap=cm.jet, interpolation='nearest')
  File "/usr/lib/pymodules/python2.7/matplotlib/axes.py", line 6796, in imshow
    im.set_data(X)
  File "/usr/lib/pymodules/python2.7/matplotlib/image.py", line 411, in set_data
    raise TypeError("Invalid dimensions for image data")
TypeError: Invalid dimensions for image data

Your code looks like it might have some indentation problems, which often happens when using mixed tabs and spaces. So I’d recommend (1) trying python -tt yourprogramname.py to see if there are whitespace errors, and (2) making sure that you’re using 4-space tabs throughout.