How to Calculate Median Using NumPy in Python

Use numpy.median() to calculate the median of an array or list. Pass your data as the first argument, and optionally specify the axis to calculate median along rows or columns.

Syntax

The basic syntax of numpy.median() is:

• numpy.median(a, axis=None, out=None, overwrite_input=False, keepdims=False)

Where:

• a: Input array or list of numbers.
• axis: Axis along which to compute the median. Default is None (median of flattened array).
• out: Optional output array to store result.
• overwrite_input: If True, allows modifying input array for speed.
• keepdims: If True, retains reduced dimensions with size 1.

python

numpy.median(a, axis=None, out=None, overwrite_input=False, keepdims=False)

Example

This example shows how to calculate the median of a 1D and 2D array using numpy.median(). It also demonstrates calculating median along a specific axis.

python

import numpy as np

# 1D array median
arr1d = np.array([10, 20, 30, 40, 50])
median_1d = np.median(arr1d)

# 2D array median (flattened)
arr2d = np.array([[1, 3, 5], [2, 4, 6]])
median_2d_flat = np.median(arr2d)

# Median along axis 0 (columns)
median_axis0 = np.median(arr2d, axis=0)

# Median along axis 1 (rows)
median_axis1 = np.median(arr2d, axis=1)

print(f"Median of 1D array: {median_1d}")
print(f"Median of 2D array (flattened): {median_2d_flat}")
print(f"Median along axis 0: {median_axis0}")
print(f"Median along axis 1: {median_axis1}")

Output

Median of 1D array: 30.0
Median of 2D array (flattened): 3.5
Median along axis 0: [1.5 3.5 5.5]
Median along axis 1: [3. 4.]

Common Pitfalls

Common mistakes when calculating median with NumPy include:

• Passing a list without converting to a NumPy array is allowed but slower.
• Not specifying axis when working with multi-dimensional arrays can lead to unexpected results.
• For even-length data, median is the average of the two middle values, which might be surprising.
• Using overwrite_input=True modifies the original array, which can cause bugs if you reuse the data.

python

import numpy as np

# Wrong: median of 2D array without axis might confuse beginners
arr = np.array([[1, 2], [3, 4]])
median_wrong = np.median(arr)  # median of flattened array

# Right: specify axis to get median per row or column
median_axis0 = np.median(arr, axis=0)  # median per column
median_axis1 = np.median(arr, axis=1)  # median per row

print(f"Median without axis (flattened): {median_wrong}")
print(f"Median along axis 0: {median_axis0}")
print(f"Median along axis 1: {median_axis1}")

Output

Median without axis (flattened): 2.5
Median along axis 0: [2. 3.]
Median along axis 1: [1.5 3.5]

Quick Reference

Summary tips for using numpy.median():

• Use axis=None for median of all elements.
• Use axis=0 for median of each column in 2D arrays.
• Use axis=1 for median of each row in 2D arrays.
• Input can be a list or NumPy array; arrays are faster.
• Median of even-length data is the average of middle two values.

Key Takeaways

Use numpy.median() to find the median of data arrays easily.

Specify the axis parameter to calculate median along rows or columns in multi-dimensional arrays.

Median of even-length data is the average of the two middle values.

Passing a NumPy array is faster than a list but both work.

Avoid overwrite_input=True unless you want to modify the original data.