ABOUT THIS WORKSHOP
Learn A/B Testing and Multi-Arm Bandit for Data Science
Interested in learning Data Science and Machine Learning but apprehensive about the statistics and coding requirements? This workshop will be an example of how you can confidently learn fundamental statistics and coding elements needed to understand and practice Data Science and Machine Learning. In particular, you will learn the following important concepts:
A/B testing is a classic statistical method for comparing two competing options (A, B). The Multi-Arm Bandit algorithm is a modern generalization of A/B testing, which uses a reinforcement learning approach to not just compare but also exploit multiple options (A, B, C, etc) at the same time.
We start by introducing A/B testing from “scratch”, using the fundamental concepts of resampling and permutations together with helpful graphical explanations (histogram plot). Then, we use A/B testing to introduce important general concepts in hypothesis testing, such as the p-value. Finally, we present an overview of the Multi-Arm Bandit algorithm, as a modern generalization of A/B testing, and as an example of reinforcement learning.
This introductory-level workshop is perfect for both non-programmers who do not have any programming experience and are interested in learning about data science, and programmers who are looking to brush up their skills or expand their programming toolkit. We will use a Jupyter notebook (Python) to drive this workshop with code, visualizations, and examples illustrating the important concepts.
