I have a list of numpy arrays, and want to remove elements based on a specific position.
[array([ 1, 2, nan, 4, nan]),
array([ 5, 6, 7, 8, 9]),
array([ 5, 6, 7, 8, 9])]
I need to get the positions of all the np.nan and remove the corresponding columns, giving me:
[array([ 1, 2, 4]),
array([ 5, 6, 8]),
array([ 5, 6, 8])]
Short of looping everything, I have no idea where to start!
Given that your list of arrays is named A:
A = [np.delete(row,np.argwhere(np.isnan(A[0]))) for row in A]
output:
[array([1., 2., 4.]), array([5, 6, 8]), array([5, 6, 8])]
Depending on your application however it might be better to use sams-studio's approach of creating one large numpy array. I would prefer that approach for most applications.
I guess the easiest would be
# arr is your original list
arr = [np.array([ 1, 2, np.nan, 4, np.nan]),
np.array([ 5, 6, 7, 8, 9]),
np.array([ 5, 6, 7, 8, 9])]
return_array = np.asarray(arr)[:, ~np.isnan(np.asarray(arr)).any(axis=0)]
print(return_array)
This will return:
array([[1., 2., 4.],
[5., 6., 8.],
[5., 6., 8.]])
If you want to have a list of array, then you have to do additionally:
return_array = [r for r in return_array]
