Search⌘ K
AI Features
Home
Courses
Reliable Machine Learning

Unit Testing

Explore unit testing as a crucial practice for verifying individual functions and methods in machine learning code. Understand how unit tests can document expected behavior, enhance code modularity, and catch bugs early. Learn the benefits and limitations of unit tests to effectively integrate them into broader testing strategies for more reliable ML systems.

We'll cover the following...

Overview of unit tests

Let’s start with unit tests. Here’s an example to give you an intuitive sense of what they are.

A project’s code is a complex web of transformations called functions or class methods, which transform some input into some output. Some of these transformations are deterministic, and some of them are random.

Why not check these transformations at every step, to see whether each result corresponds to the expected one in some sense (either a complete match or satisfaction of certain conditions)? These simple, function-wise tests are called unit tests, where a function refers to a unit.

Code example

Let’s look at some examples of unit tests. In the code below, we define a function, average, that computes the average of some numbers. We then test if it gives us the expected output while passing some input.

Python 3.8
def average(x):
    return sum(x) / len(x)
def test_average():
    x = [0, 1, 5]
    assert average(x) == 2

Let’s look at an example of ...