How to solve issue of PendingDeprecationWarning: the matrix subclass is not the recommended way to represent matrices

Question

I am using a python code as a black box that I do not want to touch. The code was working well using Python under Ubuntu 12.04 but after upgrading the system to Ubuntu 16, I got the below warning which interrupt the code from running. Any idea how I can fix this without changing the code? Many thanks.

File "/home/hammouc/.local/lib/python2.7/site-packages/scipy/sparse/base.py", line 849, in todense return np.asmatrix(self.toarray(order=order, out=out))

File "/home/hammouc/.local/lib/python2.7/site-packages/numpy/matrixlib/defmatrix.py", line 71, in asmatrix return matrix(data, dtype=dtype, copy=False)

File "/home/hammouc/.local/lib/python2.7/site-packages/numpy/matrixlib/defmatrix.py", line 123, in new PendingDeprecationWarning, stacklevel=2)

PendingDeprecationWarning: the matrix subclass is not the recommended way to represent matrices or deal with linear algebra (see https://docs.scipy.org/doc/numpy/user/numpy-for-matlab-users.html). Please adjust your code to use regular ndarray.

Looks like a scipy.sparse matrix todense is triggering this. The ideal is to replace this with a toarray method call. — hpaulj
Which numpy and scipy versions are you using? I know we discourage the use of np.matrix, but I wasn't aware of any actual move toward deprecation. — hpaulj
I found that warning message in the np.matrixlib.defmatrix.py file (1.16.3). Your warnings level must be different than mine (default). — hpaulj

hpaulj hpaulj · Accepted Answer · 2019-05-15T18:07:20

With the import warnings module, it's possible to control the display of warnings.

With M as a sparse matrix:

In [26]: warnings.filterwarnings('ignore', category=PendingDeprecationWarning)  
In [27]: M.todense()                                                            
Out[27]: 
matrix([[1., 0., 0.],
        [0., 1., 0.],
        [0., 0., 1.]])
In [28]: warnings.filterwarnings('default', category=PendingDeprecationWarning) 
In [29]: M.todense()                                                            
/usr/local/lib/python3.6/dist-packages/numpy/matrixlib/defmatrix.py:71: PendingDeprecationWarning: the matrix subclass is not the recommended way to represent matrices or deal with linear algebra (see https://docs.scipy.org/doc/numpy/user/numpy-for-matlab-users.html). Please adjust your code to use regular ndarray.
  return matrix(data, dtype=dtype, copy=False)
Out[29]: 
matrix([[1., 0., 0.],
        [0., 1., 0.],
        [0., 0., 1.]])

Producing a ndarray instead of a np.matrix:

In [30]: M.toarray()                                                            
Out[30]: 
array([[1., 0., 0.],
       [0., 1., 0.],
       [0., 0., 1.]])
In [31]: M.A                                                                    
Out[31]: 
array([[1., 0., 0.],
       [0., 1., 0.],
       [0., 0., 1.]])

Apparently my default ipython setup ignores these warnings, so I haven't seen this before. I'll have to look at the config file.

How to solve issue of PendingDeprecationWarning: the matrix subclass is not the recommended way to represent matrices

2 Answers