NumPy Indexing & Slicing

Why indexing & slicing matter

Indexing and slicing let you access and transform subsets of data without loops. Most slices return views (no copy), making them fast and memory-efficient.

1D indexing & slicing

import numpy as np

a = np.array([10, 20, 30, 40, 50])
print(a[0])        # 10 (first element)
print(a[-1])       # 50 (last element)

# slice: start:stop (stop is exclusive)
print(a[1:4])      # [20 30 40]
print(a[:3])       # [10 20 30]
print(a[3:])       # [40 50]

# step / stride
print(a[::2])      # [10 30 50]
print(a[::-1])     # [50 40 30 20 10]  (reverse)

2D indexing (rows, columns)

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

print(b[0, 1])     # 2  (row 0, col 1)
print(b[1])        # [10 20 30] (row 1)
print(b[:, 0])     # [ 1 10 40] (first column)

# Sub-matrix: rows 0..1, cols 1..2
print(b[0:2, 1:3]) # [[ 2  3],
                   #  [20 30]]

# Every row, every other col
print(b[:, ::2])   # [[ 1  3],
                   #  [10 30],
                   #  [40 60]]

3D indexing (example)

c = np.arange(2*3*4).reshape(2,3,4)  # shape (depth=2, rows=3, cols=4)
# Pick depth=1, rows 0..1, cols 2..3
print(c[1, 0:2, 2:4])
# Ellipsis to keep all middle dims:
print(c[..., 0])   # first column from each 2D slice

Views vs Copies (important!)

a = np.array([0, 1, 2, 3, 4, 5])
s = a[1:4]     # slice -> view (shares memory)
s[0] = 99
print(a)       # [ 0 99  2  3  4  5]  (a changed!)

# Force a copy if you need independence:
s_copy = a[1:4].copy()
s_copy[0] = -1
print(a[1:4])  # original remains unchanged now

Boolean indexing & where()

Use boolean masks to filter arrays. np.where builds masks or chooses values conditionally.

x = np.array([5, 12, 7, 22, 3, 18])
mask = x >= 10
print(mask)         # [False  True False  True False  True]
print(x[mask])      # [12 22 18]

# Conditional replace: values < 10 become 0
y = np.where(x < 10, 0, x)
print(y)            # [ 0 12  0 22  0 18]

Common patterns & pitfalls

• Stop is exclusive: a[m:n] includes m but not n.
• Negative indices: -1 is last element, -2 second last, etc.
• Strides: a[start:stop:step]; reverse with step=-1.
• Row vs column: In 2D, a[i] is a row; use a[:, j] for a column.
• Views change the original: copy with .copy() when you need isolation.

Practice: quick exercises

# 1) Extract every 3rd element from 0..29 starting at 2
a = np.arange(30)
print(a[2::3])

# 2) From a 6x6 array, get the 3x3 center block
b = np.arange(36).reshape(6,6)
print(b[1:4, 1:4])

# 3) Replace all odd numbers in 0..19 with -1 using boolean mask
c = np.arange(20)
c[c % 2 == 1] = -1
print(c)

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: