Unplugged Decision Tree Learning – A Learning Activity for Machine Learning Education in K-12

Abstract

Artificial intelligence (AI) is now deeply ingrained in young peoples’ everyday lives. They need low-threshold learning opportunities to understand what AI is and how it works. Unplugged learning activities can offer such opportunities but must first manage to break down the topic’s complexities. This contribution presents such an activity giving students hands-on experience in training an actual machine learning (ML) model - all without a computer! ‘Actual’ here refers to the fact that the model students train ends up being an exact copy of what a standard Python implementation would produce. Three tools are presented that make this feasible in an unplugged two-hour workshop setting. We report our experience piloting the activity including questionnaire responses we collected from 56 upper secondary school students.