Delete row from nd-array by value-CodePudding

I have an ND array, and I want to delete a specific row by value, How to do so? Currently, the output is an empty list.

 import numpy as np


def delete_item(array, item):
    new_a = np.delete(array, np.where(array == item), axis=0)
    return new_a


a = np.array([[1, 1, 1], [85, 32, 531], [78, 91, 79]])
new_array = delete_item(a, [1, 1, 1])  # -> it returns empty list []
"""
    The output is: 
        []
    What i want is:
        [[85, 32, 531],
         [78, 91, 79]]
"""
print(new_array)

CodePudding user response：

Here is one possibility:

def delete_item(array, item):
    new_a = array.copy()
    return new_a[(array != item).any(1)]
    # or
    # return np.delete(array, np.where((array == item).all(1))[0], axis=0)

a = np.array([[1, 1, 1], [85, 32, 531], [78, 91, 79]])
new_array = delete_item(a, [1, 1, 1])

output:

array([[ 85,  32, 531],
       [ 78,  91,  79]])

CodePudding user response：

The issue is in your comparison operator within where. This works:

    def delete_item(array, item):
        new_a = np.delete(array, np.all(array == item, axis=1), axis=0)
        return new_a

Originally, array == item gives:

[[ True  True  True]
 [False False False]
 [False False False]]

which provides an element by element comparison broadcasting [1, 1, 1] appropriately. Using np.all(array == item, axis=1)) gives:

[ True False False]

which is what you want. In fact, you do not need the np.where.