Splitting Arrays into Chunks (split, hsplit, vsplit)0%

Vectorization: Eliminating Python Loops with SIMD

Splitting Arrays into Chunks (split, hsplit, vsplit)

Beginner12 min read•Updated: Jul 11, 2026

Splitting Arrays: split(), hsplit(), & vsplit()

The reverse of array joining is array splitting—breaking a large array into multiple smaller sub-arrays. This is standard in machine learning pipelines when splitting feature matrices $X$ from label vectors $y$, or dividing a dataset into equal batches.

NumPy provides three intuitive splitting routines:

np.split(): General splitting along any specified axis.
np.vsplit(): Vertical splitting (row-wise / along axis 0).
np.hsplit(): Horizontal splitting (column-wise / along axis 1).

1. General Splitting with `np.split()`

np.split(ary, indices_or_sections, axis=0) accepts two modes for dividing an array:

1
An integer $N$: Splits the array into $N$ equal-sized sub-arrays. (The dimension size along the split axis must be evenly divisible by $N$).
2
A list of integer indices: Splits the array at specific partition boundaries.

Mode 1: Equal Sections

1import numpy as np
2 
3arr = np.arange(12)  # [0, 1, 2, ..., 11]
4 
5# Split into 3 equal parts (12 / 3 = 4 elements each)
6parts = np.split(arr, 3)
7for i, p in enumerate(parts):
8    print(f"Part {i}:", p)

Output:

1Part 0: [0 1 2 3]
2Part 1: [4 5 6 7]
3Part 2: [8 9 10 11]

If the array size is not divisible by $N$, np.split() raises ValueError: array split does not result in an equal division. To allow unequal splitting, use np.array_split().

Mode 2: Explicit Index Partitions

1# Split at index 2 and index 7
2# Results in: arr[:2], arr[2:7], arr[7:]
3sections = np.split(arr, [2, 7])
4print("Section 0 (0 to 2):", sections[0])
5print("Section 1 (2 to 7):", sections[1])
6print("Section 2 (7 to end):", sections[2])

2. Vertical Splitting: `np.vsplit()`

np.vsplit() splits a 2D array vertically (dividing rows). It is equivalent to np.split(..., axis=0):

1dataset = np.arange(24).reshape(6, 4)
2print("Original 6x4 Dataset:
3", dataset)
4 
5# Split 6 rows into 2 equal halves (3 rows each)
6top_half, bottom_half = np.vsplit(dataset, 2)
7print("
8Top half (3 rows):
9", top_half)
10print("
11Bottom half (3 rows):
12", bottom_half)

3. Horizontal Splitting: `np.hsplit()`

np.hsplit() splits an array horizontally (dividing columns). It is equivalent to np.split(..., axis=1).

Real-World Machine Learning Example: Splitting Features $X$ and Target $y$

In supervised learning datasets, the target label $y$ is often the last column, while features $X$ occupy all preceding columns:

1# 5 samples with 3 features + 1 target label (5x4 matrix)
2data = np.array([
3    [1.5, 2.3, 0.9, 1],
4    [3.1, 1.2, 4.4, 0],
5    [2.8, 5.1, 3.2, 1],
6    [0.9, 1.8, 2.1, 0],
7    [4.2, 3.9, 5.0, 1]
8])
9 
10# Split horizontally after index 3: columns 0-2 (features), column 3 (target)
11X, y = np.hsplit(data, [3])
12print("Features X (Shape:", X.shape, "):
13", X)
14print("
15Target y (Shape:", y.shape, "):
16", y)

4. Unequal Splitting with `np.array_split()`

When you need to divide a dataset of say 10 samples across 3 workers, $10$ is not divisible by $3$. np.array_split() cleanly distributes the remainder across the first sub-arrays:

1ten_items = np.arange(10)
2 
3# Splits 10 items into 3 chunks: lengths 4, 3, and 3
4chunks = np.array_split(ten_items, 3)
5for i, c in enumerate(chunks):
6    print(f"Worker {i} chunk (size {len(c)}):", c)

Output:

1Worker 0 chunk (size 4): [0 1 2 3]
2Worker 1 chunk (size 3): [4 5 6]
3Worker 2 chunk (size 3): [7 8 9]

Interactive Quiz• Knowledge Check

Multiple Choice Questions

Question 1

What happens if you call np.split(np.arange(10), 3)?

Question 2

What slices are created by np.split(arr, [3, 8])?

Question 3

Which function is equivalent to np.split(matrix, 2, axis=0) for a 2D matrix?

Question 4

Given a matrix of shape (8, 6), what is the shape of each sub-array after np.hsplit(matrix, 3)?

Question 5

How does np.array_split(arr, N) differ from np.split(arr, N)?

Next Lesson

Vectorization: Eliminating Python Loops with SIMD

Continue learning with hands-on practice, examples, and exercises in the upcoming topic.

Related Lessons

Previous Lesson	Next Lesson
Depth & Column Stacking (dstack, column_stack, stack)	Vectorization: Eliminating Python Loops with SIMD

Practice Quiz

Test your understanding of this lesson with 5 questions. Each question has one correct answer.

Previous Next

PreviousPrevNext

MSK Institute • Shikohabad

Loading Page Content...

Preparing courses, live labs, and learning resources.सामग्री लोड हो रही है, कृपया प्रतीक्षा करें...

Connecting serverPlease wait...

💻 Practical Labs🚀 Free Demo📜 Verified Certificates

Vectorization: Eliminating Python Loops with SIMD

Splitting Arrays into Chunks (split, hsplit, vsplit)

Beginner12 min read•Updated: Jul 11, 2026

Study Materials Cheatsheet

Splitting Arrays: split(), hsplit(), & vsplit()

NumPy provides three intuitive splitting routines:

np.split(): General splitting along any specified axis.
np.vsplit(): Vertical splitting (row-wise / along axis 0).
np.hsplit(): Horizontal splitting (column-wise / along axis 1).

1. General Splitting with `np.split()`

np.split(ary, indices_or_sections, axis=0) accepts two modes for dividing an array:

1
An integer $N$: Splits the array into $N$ equal-sized sub-arrays. (The dimension size along the split axis must be evenly divisible by $N$).
2
A list of integer indices: Splits the array at specific partition boundaries.

Mode 1: Equal Sections

1import numpy as np
2 
3arr = np.arange(12)  # [0, 1, 2, ..., 11]
4 
5# Split into 3 equal parts (12 / 3 = 4 elements each)
6parts = np.split(arr, 3)
7for i, p in enumerate(parts):
8    print(f"Part {i}:", p)

Output:

1Part 0: [0 1 2 3]
2Part 1: [4 5 6 7]
3Part 2: [8 9 10 11]

If the array size is not divisible by $N$, np.split() raises ValueError: array split does not result in an equal division. To allow unequal splitting, use np.array_split().

Mode 2: Explicit Index Partitions

1# Split at index 2 and index 7
2# Results in: arr[:2], arr[2:7], arr[7:]
3sections = np.split(arr, [2, 7])
4print("Section 0 (0 to 2):", sections[0])
5print("Section 1 (2 to 7):", sections[1])
6print("Section 2 (7 to end):", sections[2])

2. Vertical Splitting: `np.vsplit()`

np.vsplit() splits a 2D array vertically (dividing rows). It is equivalent to np.split(..., axis=0):

1dataset = np.arange(24).reshape(6, 4)
2print("Original 6x4 Dataset:
3", dataset)
4 
5# Split 6 rows into 2 equal halves (3 rows each)
6top_half, bottom_half = np.vsplit(dataset, 2)
7print("
8Top half (3 rows):
9", top_half)
10print("
11Bottom half (3 rows):
12", bottom_half)

3. Horizontal Splitting: `np.hsplit()`

np.hsplit() splits an array horizontally (dividing columns). It is equivalent to np.split(..., axis=1).

Real-World Machine Learning Example: Splitting Features $X$ and Target $y$

In supervised learning datasets, the target label $y$ is often the last column, while features $X$ occupy all preceding columns:

1# 5 samples with 3 features + 1 target label (5x4 matrix)
2data = np.array([
3    [1.5, 2.3, 0.9, 1],
4    [3.1, 1.2, 4.4, 0],
5    [2.8, 5.1, 3.2, 1],
6    [0.9, 1.8, 2.1, 0],
7    [4.2, 3.9, 5.0, 1]
8])
9 
10# Split horizontally after index 3: columns 0-2 (features), column 3 (target)
11X, y = np.hsplit(data, [3])
12print("Features X (Shape:", X.shape, "):
13", X)
14print("
15Target y (Shape:", y.shape, "):
16", y)

4. Unequal Splitting with `np.array_split()`

When you need to divide a dataset of say 10 samples across 3 workers, $10$ is not divisible by $3$. np.array_split() cleanly distributes the remainder across the first sub-arrays:

1ten_items = np.arange(10)
2 
3# Splits 10 items into 3 chunks: lengths 4, 3, and 3
4chunks = np.array_split(ten_items, 3)
5for i, c in enumerate(chunks):
6    print(f"Worker {i} chunk (size {len(c)}):", c)

Output:

1Worker 0 chunk (size 4): [0 1 2 3]
2Worker 1 chunk (size 3): [4 5 6]
3Worker 2 chunk (size 3): [7 8 9]

Interactive Quiz• Knowledge Check

Multiple Choice Questions

Question 1

What happens if you call np.split(np.arange(10), 3)?

Question 2

What slices are created by np.split(arr, [3, 8])?

Question 3

Which function is equivalent to np.split(matrix, 2, axis=0) for a 2D matrix?

Question 4

Given a matrix of shape (8, 6), what is the shape of each sub-array after np.hsplit(matrix, 3)?

Question 5

How does np.array_split(arr, N) differ from np.split(arr, N)?

Next Lesson

Vectorization: Eliminating Python Loops with SIMD

Continue learning with hands-on practice, examples, and exercises in the upcoming topic.

Related Lessons

Previous Lesson	Next Lesson
Depth & Column Stacking (dstack, column_stack, stack)	Vectorization: Eliminating Python Loops with SIMD

Practice Quiz

Test your understanding of this lesson with 5 questions. Each question has one correct answer.

Previous Next

Splitting Arrays: split(), hsplit(), & vsplit()

1. General Splitting with np.split()

Mode 1: Equal Sections

Mode 2: Explicit Index Partitions

2. Vertical Splitting: np.vsplit()

3. Horizontal Splitting: np.hsplit()

Real-World Machine Learning Example: Splitting Features $X$ and Target $y$

4. Unequal Splitting with np.array_split()

Multiple Choice Questions

Vectorization: Eliminating Python Loops with SIMD

Related Lessons

Practice Quiz

Loading Page Content...

Splitting Arrays: split(), hsplit(), & vsplit()

1. General Splitting with np.split()

Mode 1: Equal Sections

Mode 2: Explicit Index Partitions

2. Vertical Splitting: np.vsplit()

3. Horizontal Splitting: np.hsplit()

Real-World Machine Learning Example: Splitting Features $X$ and Target $y$

4. Unequal Splitting with np.array_split()

Multiple Choice Questions

Vectorization: Eliminating Python Loops with SIMD

Related Lessons

Practice Quiz

1. General Splitting with `np.split()`

2. Vertical Splitting: `np.vsplit()`

3. Horizontal Splitting: `np.hsplit()`

4. Unequal Splitting with `np.array_split()`

1. General Splitting with `np.split()`

2. Vertical Splitting: `np.vsplit()`

3. Horizontal Splitting: `np.hsplit()`

4. Unequal Splitting with `np.array_split()`