Tim Erickson (Epistemological Engineering)
Abstract
Bring your laptops! You’ll use CODAP and disease data to make classification trees—not by using an algorithm (such as CART), but “by hand,” using common sense and seat-of-the-pants modeling. How do decision points in a tree relate to traditional data plots? How can we use data to make a tree as accurate as possible? What does accuracy even mean in this context? We’ll see how students can grasp the principles underlying tree algorithms, and open a window into machine learning. No experience necessary. You’ll learn the basics of CODAP as we go—these basics will only take you a minute or two—as well as more sophisticated CODAP techniques such as how to use hierarchical data organization. Our principal tool will be the “Arbor” plugin for making trees. And everything will be online so you can come back to it later. Did we mention that CODAP is free? Always and forever. And there’s a game! You get to be a xenobiologist helping diagnose alien diseases.