[SciPy-user] Error using 0.4.8 on code that worked on 0.3.2

[Antonio Arauzo Azofra]

Hello everybody,

I am almost new to Scipy. I was using the following in my code on 0.3.2 
and it worked:

 >>> mat=[[1,2,3],[4,5,6],[7,8,9]]
 >>> import scipy
 >>> scipy.cov(mat)
Traceback (most recent call last):
   File "<stdin>", line 1, in ?
   File "/usr/lib/python2.4/site-packages/numpy/lib/function_base.py", 
line 649, in cov
     if (m.shape[0] == 1):
AttributeError: 'list' object has no attribute 'shape'

Now it is necessary to use always an array:

 >>> amat = array(mat)
Traceback (most recent call last):
   File "<stdin>", line 1, in ?
NameError: name 'array' is not defined
 >>> amat = scipy.array(mat)
 >>> scipy.cov(amat)
array([[ 9.,  9.,  9.],
        [ 9.,  9.,  9.],
        [ 9.,  9.,  9.]])

Is this better? If it is, (by no means I try to critize good work made 
on scipy) maybe it would be better if 'cov' funtion checked the type 
before begining, and throw an more meaningfull error. :-?  Just a 
suggestion.

Also I had problems because cov function returns an integer if matrix is 
1x1. I have used this workarround but I am not sure if this is the best 
way to work with Scipy. Any suggestions?

-----------------------------------------------------------
     ...
     disp1 = scipy.cov(class1) #within class1 scatter matrix
     disp2 = scipy.cov(class2) #within class2 scatter matrix

     #We allways want covariance matrices even if they are 1 by 1
     if not type(disp1) == type( scipy.array(0) ):
        disp1 = scipy.array([[disp1]])
     if not type(disp2) == type( scipy.array(0) ):
        disp2 = scipy.array([[disp2]])

     aux = scipy.add(disp1,disp2)
     aux = scipy.divide(aux,2)
     ...
-----------------------------------------------------------

-- 
Regards,
   Antonio Arauzo Azofra