Copy vs View in NumPy

When working with NumPy arrays, understanding how data is stored in memory is extremely important. Some operations create a copy of data, while others create a view that shares the same memory.

This lesson explains the difference clearly, with real examples and outputs.

Why Copy vs View Matters

Knowing whether an operation creates a copy or a view helps you:

• Avoid unexpected data changes
• Improve performance and memory usage
• Write safer and more efficient code
• Debug data-related issues faster

Original Array

Let’s start with a simple NumPy array.

import numpy as np

data = np.array([10, 20, 30, 40, 50])
print(data)

Output:

[10 20 30 40 50]

What Is a View?

A view is a new array object that looks at the same underlying data in memory. Changing a view also changes the original array.

view_array = data.view()
view_array[0] = 999

print("View:", view_array)
print("Original:", data)

Output:

View: [999  20  30  40  50]
Original: [999  20  30  40  50]

Both arrays changed because they share the same memory.

Checking Memory Sharing

You can check whether arrays share memory using np.shares_memory().

np.shares_memory(data, view_array)

Output:

True

What Is a Copy?

A copy creates a completely new array with its own memory. Changes to the copy do not affect the original array.

copy_array = data.copy()
copy_array[1] = 777

print("Copy:", copy_array)
print("Original:", data)

Output:

Copy: [999 777  30  40  50]
Original: [999  20  30  40  50]

The original array remains unchanged.

Memory Check for Copy

np.shares_memory(data, copy_array)

Output:

False

Common Operations That Create Views

• Slicing arrays
• Using view()
• Reshaping (in many cases)

Example with slicing:

sliced = data[1:4]
sliced[0] = 111

print("Sliced:", sliced)
print("Original:", data)

Common Operations That Create Copies

• copy()
• Advanced indexing
• Some mathematical operations

When to Use Copy vs View

• Use view when performance and memory efficiency matter
• Use copy when you want data safety
• Always be cautious when modifying sliced arrays

Practice Exercise

Exercise

Create a NumPy array and:

• Create a view and modify it
• Create a copy and modify it
• Observe how the original array changes

Expected Outcome

You should clearly understand how memory sharing works in NumPy.

What’s Next?

In the next lesson, you will learn about broadcasting and how NumPy performs operations on arrays with different shapes.