NumPy Boolean Masks & Conditional Selection

Boolean indexing evaluates element-wise relational operations (such as >, ==, <=) to produce Boolean arrays. These masks can be combined using bitwise operators (&, |, ~) or logical functions like np.logical_and. For conditional mapping, see where().

What is a Boolean Mask? 🔝

A boolean mask is an array of truth values matching the shape of the target array. When passed inside indexing brackets x[mask], NumPy extracts only the positions where the mask evaluates to True.

Filter 1D Arrays 🔝

Create relational conditions and filter elements:

import numpy as np

x = np.array([5, 12, 7, 22, 3, 18])

mask = x >= 10
print("Mask:", mask)         # [False  True False  True False  True]
print("Filtered:", x[mask])  # [12 22 18]

# Combine conditions (use parentheses with & and |)
mask2 = (x >= 10) & (x % 2 == 0)
print("Combined condition filtered:", x[mask2]) # [12 22 18]

assert np.array_equal(x[mask2], np.array([12, 22, 18]))

Expected output

Mask: [False  True False  True False  True]
Filtered: [12 22 18]
Combined condition filtered: [12 22 18]

Conditional Replace with np.where 🔝

Use np.where(cond, a_if_true, a_if_false) to construct new arrays based on conditions:

x = np.array([5, 12, 7, 22, 3, 18])

# Replace values < 10 with 0, keep others
y = np.where(x < 10, 0, x)
print("Replaced array:", y)  # [ 0 12  0 22  0 18]

# Map categories
labels = np.where(x >= 15, 'high', 'low')
print("Category labels:", labels)  # ['low' 'low' 'low' 'high' 'low' 'high']

Expected output

Replaced array: [ 0 12  0 22  0 18]
Category labels: ['low' 'low' 'low' 'high' 'low' 'high']

Logical Functions vs Operators 🔝

Bitwise operators require parentheses due to Python operator precedence rules. Alternatively, use logical functions:

a = np.arange(12)
m1 = (a >= 3) & (a <= 8)               # Operators
m2 = np.logical_and(a >= 3, a <= 8)     # Functions

print(np.array_equal(a[m1], a[m2])) # True
assert np.array_equal(a[m1], np.array([3, 4, 5, 6, 7, 8]))

Expected output

True

Masking 2D Arrays 🔝

Apply masks along rows or element-wise across 2D matrices:

b = np.array([
    [1, 2, 3],
    [10, 20, 30],
    [40, 50, 60]
])

# Column-wise filter: take rows where column 0 >= 10
row_mask = b[:, 0] >= 10
print("Filtered rows:
", b[row_mask])

# Element-wise mask: extract 1D array of elements >= 20
elem_mask = b >= 20
print("Filtered elements:", b[elem_mask])

assert b[row_mask].shape == (2, 3)

Expected output

Filtered rows:
 [[10 20 30]
 [40 50 60]]
Filtered elements: [20 30 40 50 60]

NaN-Safe Filters 🔝

Use np.isfinite, np.isnan, and np.isinf to handle special floating-point values safely:

c = np.array([1.0, np.nan, 3.5, np.inf, -np.inf, 2.2])

finite_mask = np.isfinite(c)
print("Finite elements:", c[finite_mask])

# Replace NaNs with mean of finite values
c_fixed = np.where(np.isnan(c), np.nanmean(c), c)
print("NaN-fixed array:", c_fixed)

assert np.array_equal(c[finite_mask], np.array([1.0, 3.5, 2.2]))

Expected output

Finite elements: [1.  3.5 2.2]
NaN-fixed array: [ 1.25  2.25  3.5   inf  -inf   2.2  ]

In-Place Updates with Masks 🔝

Modify elements matching a boolean condition directly in place:

d = np.arange(10)
d[d % 2 == 1] = -1
print("Updated array:", d)
assert np.array_equal(d, np.array([0, -1, 2, -1, 4, -1, 6, -1, 8, -1]))

Expected output

Updated array: [ 0 -1  2 -1  4 -1  6 -1  8 -1]

Practice in Google Colab 🔝

Test boolean indexing, conditional mapping with np.where, 2D masking, and run verification test suites in our interactive Google Colab companion notebook:

Interactive Practice: Colab & GitHub

Launch the notebook directly in Google Colab to run live code cells or inspect the source repository on GitHub:

View on GitHub

Continue exploring related array indexing and filtering functions: Indexing & Slicing, where(), and array().

Subhendu Mohapatra

Author & Instructor at plus2net

I write and maintain practical tutorials on Python, PHP, SQL, JavaScript, HTML, jQuery, and web development at plus2net. The tutorials focus on clear explanations, working examples, and code that readers can test and adapt while learning.

About the author   Watch on YouTube

Connect: