无法访问 Python 二维数组中的行元素？

Question

I have created a 2d matrix using Scipy's coo_matrix, and have a matrix M as such:

df = pd.DataFrame(columns=["hub", "auth", "weight"])
M = coo_matrix((df.iloc[:,2], (df.iloc[:,0],df.iloc[:,1])), shape=(len(hubs) + len(auths), len(hubs) + len(auths)))
M = M.todense()

[[0 0 0 1 1 1 0]
[0 0 0 1 1 0 0]
[0 0 0 0 0 0 1]
[0 0 0 0 0 0 0]
[0 0 0 0 0 0 0]
[0 0 0 0 0 0 0]
[0 0 0 0 0 0 0]]

I can successfully slice the array to get its columns and the elements in each column:

col = M[:,3]
val = col[0]

where val is equal to 1. I try to do something similar to extract a row:

row = M[0]
val = row[2]

which should also return 1, but instead val returns

[[0 0 0 1 1 1 0]]

What am I doing wrong here?

Answer 1

Since it is a numpy array (as DYZ pointed it that .todense() is called on the original coo_matrix):

Notice that your original matrix, or 2d array is 7 x 7 (7 rows by 7 columns). When you call col = M[:,3] , you are saying you want the 3rd column and all rows, which is a resulting 7 x 1 matrix (7 rows by 1 column). When you call col[2] , you are actually calling col[2,:] or getting the 2nd row (which is now just a 1 x 1 matrix).

Now, if you call row = M[0] , you are actually calling row = M[0,:] or getting the 0th row and all columns, which is a 1 x 7 matrix (1 column by 7 rows). Thus calling val = row[2] gives an indexerror as you only have 1 row in your new matrix. You could instead call val = row[:,2] to get the 2nd column.