Sort values of matrix in python

Question

What I have so far is:

dict={'A':[1,2,3], 'B':[2,5,4], 'C':[2,1,8]}
N=len(keys)
m=numpy.zeros(N,N)
for i in range(N):
    for j in range(N):
         m[i-1,j-1]=covariance(values[i-1],values[j-1])
         m[j-1,i-1]=covariance(values[j-1],values[i-1])
m=numpy.triu(m)

which gives me:

1   0.639  0.07
0     1    0.51
0     0      1

I dont have the column names or the row names yet. I would like something like this:

       A     B      C
A      1   0.639  0.07
B      0     1    0.51
C      0     0      1

Given this matrix, I would like to sort it in descending order by the value of the matrix so the output I would like is:

A & A: 1
B & B: 1
C & C: 1
A & B: 0.639
B & C: 0.51
A & C: 0.07
B & A: 0 #etc

From the output would like to save it into a csv file where the first column are the names and the second column are the corresponding scores

Thanks for reading.

Answer 1

Call np.sort with the axis keyword argument set to None, then reverse it with slicing:

>>> a = np.array([[1, 0.639, 0.07], [0, 1, 0.51], [0, 0, 1]])
>>> a
array([[ 1.   ,  0.639,  0.07 ],
       [ 0.   ,  1.   ,  0.51 ],
       [ 0.   ,  0.   ,  1.   ]])
>>> np.sort(a, axis=None)[::-1]
array([ 1.   ,  1.   ,  1.   ,  0.639,  0.51 ,  0.07 ,  0.   ,  0.   ,  0.   ])

If you want to know where each value is coming from, then first use np.argsort, then unravel the flattened indices:

>>> idx = np.argsort(a, axis=None)[::-1]
>>> rows, cols = np.unravel_index(idx, a.shape)
>>> a_sorted = a[rows, cols]
>>> for r, c, v in zip(rows, cols, a_sorted):
...     print 'ABC'[r], '&', 'ABC'[c], ':', v
... 
C & C : 1.0
B & B : 1.0
A & A : 1.0
A & B : 0.639
B & C : 0.51
A & C : 0.07
C & B : 0.0
C & A : 0.0
B & A : 0.0