user3396592 user3396592 - 3 months ago 14
Python Question

Remove tuples in a 2D numpy array that satisfy 2 conditions

So I have a numpy array of tuples and I want to remove all tuples where the first value is less than 0 or the second element is greater than a number, n.

So if n = 10 and we had this array:

[[-1, 5], [3, 11], [-4, 20]]


It would become this:

[[]]


I'm guessing I need to use
np.delete
and
np.where
in a smart way?

Thanks in advance.

Answer

You can solve this using logical indexing. I'm guessing in your example you meant that

[[-1, 5], [3, 11], [-4, 20]]

would become

[]

Since you said that the conditions are:

I want to remove all tuples where the first value is less than 0 or the second element is greater than a number, n.

>>> import numpy as np
>>> arr = np.array([[-1, 5], [3, 11], [-4, 20]])
>>> arr[~((arr[:,0] < 0) | (arr[:,1] > 10))]
array([], shape=(0, 2), dtype=int64)

Basically it's all down to expressing your logical requirement as the combination (through | and &) of different logical masks.

The ~ takes the inverse of the mask, since you said you wanted to delete elements that match your criteria. A better example of this method working is this:

>>> import numpy as np
>>> arr = np.array([[-1, 5], [1, 1], [3, 11], [-4, 20], [2, 9]])
>>> arr[~((arr[:,0] < 0) | (arr[:,1] > 10))]
array([[1, 1], [2, 9]])