Adam Adam - 2 months ago 6x
Python Question

Efficient ndarray operations

I'm converting some Matlab code in Python. I need to do some matrix manipulation. My matrix (A) is (right now) a 65x3 matrix. However, the number of rows is variable depending on what step I'm at in the program.

In Matlab, the code I'm working on is:

output = inv(A'*A) * A';

The following Python code reproduces the expected output just fine. I'm just curious if there is a better (more Pythonic, faster, etc) way to do this? I'm trying to stick only to basic Python and numpy.

output =, A)), np.transpose(A))

Thanks to anyone who is willing to help.


You can use a the T attribute (transposes the array). Also, if using Python 3.5, you can use @ for the dot product (see PEP 465 for details).

output = np.linalg.inv(A.T @ A) @ A.T