Zindi Error Metric Series: How to use Accuracy as an evaluation metric for machine learning
Getting started · 2 Jun 2022, 07:01

When working on a machine learning project, choosing the right error or evaluation metric is critical. This is a measure of how well your model performs at the task you built it for, and choosing the correct metric for the model is a critical task for any machine learning engineer or data scientist. Accuracy is used as an error or evaluation metric when True Positives and True Negatives are the most important and when the data is well-distributed.

How to use evaluation metrics

For Zindi competitions, we choose the evaluation metric for each competition based on what we want the model to achieve. Understanding each metric and the type of model you use each for is one of the first steps towards mastery of machine learning techniques.

Accuracy as an evaluation metric

When learning about machine learning, the easiest way to explain how well your model is performing, or how ‘correct’ you are, is to explain how accurate the model is.

The higher your result, the better.

As you progress in your machine learning journey, you will learn different ways to evaluate your solutions and more appropriate ways of reporting the success of your solution. Different error metrics take into account factors like the balance of the data (i.e. the target distribution) and the type of problem.

Zindi’s Accuracy error metric is very straightforward, it matches your submitted IDs with Zindi’s target (or answers) on the backend, and reports the percentage of these predictions you got correct.

If Zindi has chosen Accuracy as the error metric for a competition, it doesn’t mean that this is the only method you can use to evaluate your model locally. Another way you can evaluate and visualise your predictions is through a confusion matrix.

With this knowledge, you should be well equipped to use Accuracy for your next machine learning project.

Why don’t you test out your new knowledge on one of our competitions that uses Accuracy as its evaluation metric? We suggest the Gender-Based Violence Tweet Classification Challenge.