A Quick Intro Guide to A/B Testing for Data Scientists
Benefits and steps to validate your machine learning model.
Table of Contents
- Introduction
- Benefits
- Steps
- Summary
- References
Introduction
This article is aimed at informing not only data scientists but similar roles, like product managers or data analysts as well, on the benefits of AB testing and to serve as a simple guide to help you start your first AB test. First, we can define AB testing as an experiment where there is a control and variant sample compared to see how production data is impacted before fully rolling out a variant. Even if an AB test fails, AKA, the variant is worse than the control, you still are learning valuable lessons. With that being said, keep on reading if you would like to gain some actionable steps for AB testing, an incredibly useful skill for data scientists.
Benefits
Before we outline some steps with examples, we want to know why AB testing is important.
1. You lower the risk of releasing a harmful change to your product and business:
