How to Use std in NumPy: Calculate Standard Deviation Easily

Use numpy.std() to calculate the standard deviation of array elements. You can specify the axis to compute along rows or columns and control whether to use population or sample standard deviation with the ddof parameter.

Syntax

The basic syntax of numpy.std() is:

• numpy.std(a, axis=None, dtype=None, out=None, ddof=0, keepdims=False)

Where:

• a: Input array.
• axis: Axis or axes along which to compute the standard deviation. Default is None (compute over the whole array).
• dtype: Data type for computation (optional).
• out: Alternative output array to place the result (optional).
• ddof: Delta degrees of freedom. The divisor used in calculation is N - ddof. Use ddof=1 for sample standard deviation.
• keepdims: If True, retains reduced dimensions with size 1.

python

numpy.std(a, axis=None, dtype=None, out=None, ddof=0, keepdims=False)

Example

This example shows how to calculate the standard deviation of a 2D array along different axes and how ddof affects the result.

python

import numpy as np

# Create a 2D array
arr = np.array([[1, 2, 3], [4, 5, 6]])

# Standard deviation of all elements (population std)
std_all = np.std(arr)

# Standard deviation along columns (axis=0)
std_cols = np.std(arr, axis=0)

# Standard deviation along rows (axis=1)
std_rows = np.std(arr, axis=1)

# Sample standard deviation (ddof=1) of all elements
std_sample = np.std(arr, ddof=1)

print(f"Std of all elements: {std_all}")
print(f"Std along columns: {std_cols}")
print(f"Std along rows: {std_rows}")
print(f"Sample std of all elements: {std_sample}")

Output

Std of all elements: 1.707825127659933
Std along columns: [1.5 1.5 1.5]
Std along rows: [0.81649658 0.81649658]
Sample std of all elements: 1.8257418583505538

Common Pitfalls

Common mistakes when using numpy.std() include:

• Not setting ddof=1 when you want the sample standard deviation, which can lead to underestimating variability.
• Forgetting to specify axis when working with multi-dimensional arrays, resulting in a single value instead of per-row or per-column std.
• Passing integer arrays without specifying dtype=float, which can cause unexpected integer division in older NumPy versions (rare now).

python

import numpy as np

arr = np.array([[1, 2, 3], [4, 5, 6]])

# Wrong: sample std but ddof=0 (default)
wrong_std = np.std(arr)

# Right: sample std with ddof=1
right_std = np.std(arr, ddof=1)

print(f"Wrong sample std (ddof=0): {wrong_std}")
print(f"Correct sample std (ddof=1): {right_std}")

Output

Wrong sample std (ddof=0): 1.707825127659933
Correct sample std (ddof=1): 1.8257418583505538

Quick Reference

Remember these tips when using numpy.std():

• Use axis to control the dimension of calculation.
• Set ddof=1 for sample standard deviation.
• Default ddof=0 calculates population standard deviation.
• keepdims=True keeps the output shape compatible for broadcasting.

Key Takeaways

Use numpy.std() to calculate standard deviation of array elements easily.

Set axis parameter to compute std along rows, columns, or entire array.

Use ddof=1 for sample standard deviation, ddof=0 for population std.

For multi-dimensional arrays, specifying axis avoids confusion.

Keepdims=True helps maintain array shape after reduction.