numpy.cumsum() in Python — Compute Cumulative Sums with Clear Examples

Overview: numpy.cumsum() in Python

The numpy.cumsum() function returns the cumulative sum of array elements. It’s a common tool for aggregating values across an array or along a specific axis, and it works with NumPy ndarrays and array-like inputs such as nested lists. This guide shows syntax, practical examples, and tips for using numpy.cumsum effectively in Python.

Syntax

Function signature:

cumsum(array, axis=None, dtype=None, out=None)

Parameters:

• array — Input ndarray or array-like (for example, nested lists).
• axis — The axis along which the cumulative sum is computed. If None (default), the input is flattened first and the cumulative sum is computed on the flattened array.
• dtype — Optional output data type (for example float or int); useful to prevent overflow or to convert integers to floats.
• out — Optional output array to store the result.

Examples: Compute Cumulative Sums

The examples below demonstrate common usage patterns for numpy.cumsum(), including use without an axis, along axes, and with a specified data type.

1. Cumulative sum without an axis (flattened)

import numpy as np

array1 = np.array(
    [[1, 2],
     [3, 4],
     [5, 6]])

total = np.cumsum(array1)
print(f'Cumulative Sum of all the elements is {total}')

Output: [ 1 3 6 10 15 21]. The array is flattened to [1, 2, 3, 4, 5, 6] and the cumulative sum is calculated sequentially.

2. Cumulative sum along a specific axis

import numpy as np

array1 = np.array(
    [[1, 2],
     [3, 4],
     [5, 6]])

# Cumulative sum down each column (axis=0)
total_0_axis = np.cumsum(array1, axis=0)
print(f'Cumulative Sum of elements at 0-axis is:\n{total_0_axis}')

# Cumulative sum across each row (axis=1)
total_1_axis = np.cumsum(array1, axis=1)
print(f'Cumulative Sum of elements at 1-axis is:\n{total_1_axis}')

Example output:

[[ 1  2]
 [ 4  6]
 [ 9 12]]

[[ 1  3]
 [ 3  7]
 [ 5 11]]

Use axis=0 to accumulate down columns and axis=1 to accumulate across rows.

3. Specify the output data type (dtype)

import numpy as np

array1 = np.array(
    [[1, 2],
     [3, 4],
     [5, 6]])

total_1_axis = np.cumsum(array1, axis=1, dtype=float)
print(f'Cumulative Sum of elements at 1-axis is:\n{total_1_axis}')

Output:

[[ 1.  3.]
 [ 3.  7.]
 [ 5. 11.]]

Specifying dtype=float forces floating-point accumulation, which helps avoid integer overflow and preserves fractional results when needed.

Tips and Common Pitfalls

• Remember that axis=None flattens the input first; set an explicit axis to preserve the array shape during accumulation.
• Choose dtype carefully to avoid overflow for large integer arrays.
• Use the out parameter to write results into a preallocated array when memory or performance is a concern.

Further reading

For full details and edge cases, consult the official NumPy cumsum API documentation.