When working with numerical data, rounding is often necessary to control precision or present cleaner results. In Python, the NumPy library provides a suite of ufuncs (universal functions) specifically designed to perform decimal rounding efficiently on arrays.

In this article, you'll learn:

• ✅ What rounding ufuncs are in NumPy

• Differences between each rounding method

• How to use each rounding function with examples

• Full code example

• Tips and Common pitfalls

What is a ufunc in NumPy?

A ufunc (universal function) in NumPy operates element-wise on ndarray objects. They are:

• Fast – implemented in C under the hood

• Broadcast-compatible

• Vectorized – no explicit loops needed

For rounding, NumPy includes several built-in ufuncs to suit different use cases.

➕ NumPy Rounding Functions

Step-by-Step Examples

Let's look at how to use each of these rounding ufuncs with practical examples.

Setup

import numpy as np

arr = np.array([1.5, 2.3, -1.7, -2.9])

1. np.around() (or np.round_())

Rounds to nearest integer by default. You can also round to specific decimal places.

print(np.around(arr))
# Output: [ 2.  2. -2. -3.]

print(np.around(arr, decimals=1))
# Output: [ 1.5  2.3 -1.7 -2.9]

2. np.floor()

Rounds down (toward negative infinity):

print(np.floor(arr))
# Output: [ 1.  2. -2. -3.]

3. np.ceil()

Rounds up (toward positive infinity):

print(np.ceil(arr))
# Output: [ 2.  3. -1. -2.]

4. np.trunc()

Truncates decimals (toward zero):

print(np.trunc(arr))
# Output: [ 1.  2. -1. -2.]

5. np.fix()

Rounds toward zero (very similar to trunc()):

print(np.fix(arr))
# Output: [ 1.  2. -1. -2.]

✅ Full Working Example

import numpy as np

arr = np.array([3.6, 2.1, -1.5, -2.8])

print("Original:", arr)
print("Around (0 decimals):", np.around(arr))
print("Round to 1 decimal:", np.around(arr, decimals=1))
print("Floor:", np.floor(arr))
print("Ceil:", np.ceil(arr))
print("Trunc:", np.trunc(arr))
print("Fix:", np.fix(arr))

Output:

Original: [ 3.6  2.1 -1.5 -2.8]
Around (0 decimals): [ 4.  2. -2. -3.]
Round to 1 decimal: [ 3.6  2.1 -1.5 -2.8]
Floor: [ 3.  2. -2. -3.]
Ceil: [ 4.  3. -1. -2.]
Trunc: [ 3.  2. -1. -2.]
Fix: [ 3.  2. -1. -2.]

Tips

• Performance: These functions are vectorized—no need to loop through arrays.

• In-place Rounding: You can round and assign back using slicing:

  arr[:] = np.around(arr, 2)
  

• Decimals Argument: np.around() lets you round to n decimal places.

Common Pitfalls

Conclusion

NumPy’s built-in rounding ufuncs make working with numeric precision simple, fast, and expressive. Whether you're rounding to integers, floors, ceilings, or truncating decimals, NumPy provides efficient tools for every use case.

By using ufuncs like np.around(), np.floor(), np.ceil(), and np.trunc(), you avoid loops and unlock the full power of NumPy’s vectorized operations.

What's Next?

• NumPy Arithmetic Operations

• NumPy Statistical ufuncs

• NumPy Custom ufuncs