NumPy argwhere()

import numpy as np

originalArray = np.array([1, 0, 0, 4, -5])

# return the indices of elements that are not zero as a 2D array
result = np.argwhere(originalArray )

print(result)

'''
Output:  
[[0]
 [3]
 [4]]
'''

argwhere() Syntax

The syntax of argwhere() is:

numpy.argwhere(array)

argwhere() Argument

The argwhere() method takes one argument:

• array - an array whose non-zero indices are to be found

argwhere() Return Value

The argwhere() method returns indices of elements that are non-zero as a 2D array.

Example 1: numpy.argwhere() With Arrays

import numpy as np

numberArray = np.array([1, 0, 0, 4, -5])
stringArray = np.array(['Apple', 'Ball', '', 'Dog'])

# return indices of non-zero elements in numberArray as a 2D array
numberResult = np.argwhere(numberArray)
 
# return indices of non-empty elements in stringArray as a 2D array
stringResult = np.argwhere(stringArray)

print('Array of non-empty indices in numberArray:\n', numberResult)   
print('\nArray of non-empty indices in  stringArray:\n', stringResult)    

Output

Array of non-empty indices in numberArray:
[[0]
 [3]
 [4]]

Array of non-empty indices in  stringArray:
 [[0]
 [1]
 [3]]

Example 2: numpy.argwhere() With 2-D Arrays

import numpy as np

array = np.array([[1, 0, 3],
                              [2, 0, 0],
                              [0, 4, 5]])

# return indices of elements that are not zero
result = np.argwhere(array)

print(result)

Output

[[0 0]
 [0 2]
 [1 0]
 [2 1]
 [2 2]] 

Here, the output represents the positions of non-zero elements in the row-column format.

The first non-zero element is 1, which is in index [0, 0] in row-column format. Similarly, the second non-zero element is 3, which is in index [0, 2] in row-column format, and so on.

Example 3: numpy.argwhere() With Condition

We can also use argwhere() to find the indices of elements that satisfy the given condition.

import numpy as np

array = np.array([1, 2, 3, 4, 5, 6])

# check array elements for odd/even condition 
# return true if the array element is even
result = np.argwhere(array%2==0)

print(result)

Output

[[1]
 [3]
 [5]]

Note: To group the indices by the dimension, rather than element, we use nonzero().