numpy.where() in Python (original) (raw)

Last Updated : 30 Sep, 2025

numpy.where() is used for conditional selection and replacement in NumPy arrays. It can be used to:

• Find indices that satisfy a condition
• Build a new array by choosing values from two options depending on a condition.

In this example, np.where() is used with only a condition to get the indices where elements are greater than 20.

Python `

import numpy as np arr = np.array([10, 15, 20, 25, 30]) result = np.where(arr > 20) print(result)

`

**Explanation:

• **arr > 20 produces a boolean mask: [False, False, False, True, True].
• **np.where(...) with a single argument returns a tuple containing the indices of True entries.
• The returned value ****(array([3, 4]),)** indicates elements at positions 3 and 4 satisfy the condition.

Syntax:

numpy.where(condition[, x, y])

**Parameters:

• **condition: boolean array (or expression producing one).
• **x (optional): array-like or scalar used where condition is True.
• **y (optional): array-like or scalar used where condition is False.

**Return value:

• **With only condition: returns a tuple of index arrays (one per dimension).
• **With x and y: returns a new array choosing from x where True, y where False (supports broadcasting and dtype rules).

Uses of numpy.where()

1. Find indices that satisfy a condition

In this example, numpy.where() checks where the condition arr % 2 == 0 is true and returns the indices.

Python `

import numpy as np arr = np.array([2, 7, 8, 3, 10]) idx = np.where(arr % 2 == 0)
print(idx) print("Even elements:", arr[idx])

`

Output

(array([0, 2, 4]),) Even elements: [ 2 8 10]

**Explanation:

• **arr % 2 == 0 creates a boolean mask for even numbers.
• **np.where(...) returns indices of True values: (array([0,2,4]),).
• **arr[idx] uses those indices to extract the even elements.

2. Binary replacement with scalars (x and y)

By providing x and y as arguments, you can use numpy.where() to return different values depending on whether condition is true or false.

Here, numpy.where() function checks the condition arr > 20.

Python `

import numpy as np arr = np.array([10, 15, 20, 25, 30]) result = np.where(arr > 20, 1, 0) print(result)

`

**Explanation:

• **arr > 20 builds the boolean mask.
• **np.where(mask, 1, 0) returns an array with 1 where mask is True, else 0.
• **Result is an integer array representing the binary condition.

3. Pick from two arrays elementwise

In this example, for elements where condition arr1 > 20 is true, corresponding element from arr1 is chosen. Otherwise, element from arr2 is selected.

Python `

import numpy as np arr1 = np.array([10, 15, 20, 25, 30]) arr2 = np.array([100, 150, 200, 250, 300]) result = np.where(arr1 > 20, arr1, arr2) print(result)

`

Output

[100 150 200 25 30]

**Explanation:

• **arr1 > 20 identifies where to take values from **arr1 (True for indices 3 and 4).
• **np.where(condition, arr1, arr2) returns values from **arr1 where **True, otherwise from **arr2.
• Final array picks arr2 values where condition is False (indices 0–2) and arr1 values where condition is True (indices 3–4).

4. Replace negatives with zero

This example replaces negative numbers in a 2-D array with 0.

Python `

import numpy as np mat = np.array([[ 5, -2, 3], [-1, 4, -6]]) result = np.where(mat < 0, 0, mat) print(result)

`

**Explanation:

• **mat < 0 makes a boolean matrix of the same shape.
• **np.where(..., 0, mat) returns a new array where negatives are replaced by 0, others kept.
• The function broadcasts scalars (0) across the array shape.