How to Save Array to File in NumPy: Simple Guide

You can save a NumPy array to a file using np.save('filename.npy', array) for binary format or np.savetxt('filename.txt', array) for text format. These functions store the array data so you can load it later with np.load or read the text file.

Syntax

np.save(filename, array): Saves the array in a binary .npy file for efficient storage and fast loading.

np.savetxt(filename, array): Saves the array in a human-readable text file, useful for simple data or sharing.

filename: String with the file name or path.

array: The NumPy array you want to save.

python

import numpy as np

# Save array in binary format
np.save('data.npy', np.array([1, 2, 3]))

# Save array in text format
np.savetxt('data.txt', np.array([[1, 2], [3, 4]]))

Example

This example shows how to save a NumPy array to a binary file and a text file, then load them back to verify the data.

python

import numpy as np

# Create an array
arr = np.array([[10, 20, 30], [40, 50, 60]])

# Save as binary file
np.save('array.npy', arr)

# Save as text file
np.savetxt('array.txt', arr, fmt='%d')

# Load binary file
loaded_arr = np.load('array.npy')

# Load text file
loaded_txt = np.loadtxt('array.txt', dtype=int)

print('Loaded from .npy file:')
print(loaded_arr)
print('\nLoaded from .txt file:')
print(loaded_txt)

Output

Loaded from .npy file:
[[10 20 30]
 [40 50 60]]

Loaded from .txt file:
[[10 20 30]
 [40 50 60]]

Common Pitfalls

• Using np.savetxt for arrays with more than 2 dimensions will cause errors because it only supports 1D or 2D arrays.
• Saving with np.save adds a .npy extension automatically if not provided; forgetting this can cause confusion when loading.
• Text files saved with np.savetxt lose some precision and metadata compared to binary files.

python

import numpy as np

arr_3d = np.ones((2, 2, 2))

# Wrong: np.savetxt does not support 3D arrays
# np.savetxt('3d.txt', arr_3d)  # This will raise an error

# Right: Use np.save for 3D arrays
np.save('3d.npy', arr_3d)

# Loading works with np.load
loaded_3d = np.load('3d.npy')
print(loaded_3d)

Output

[[[1. 1.]
  [1. 1.]]

 [[1. 1.]
  [1. 1.]]]

Quick Reference

Function	Purpose	File Format	Supports Multi-D Arrays
np.save(filename, array)	Save array in binary format	.npy	Yes
np.savetxt(filename, array)	Save array in text format	.txt or .csv	Only 1D or 2D

Key Takeaways

Use np.save to store arrays in efficient binary .npy files for any shape.

Use np.savetxt to save arrays as readable text files but only for 1D or 2D arrays.

Always load saved arrays with np.load for .npy files and np.loadtxt for text files.

Remember np.save adds .npy extension automatically if missing.

Avoid np.savetxt for arrays with more than two dimensions to prevent errors.