Visual Analytic Workflow to Understand Students’ Performance in Computer Science Courses

Abstract

This Work in Progress Research Paper presents a visual analytic workflow to assist instructors of introductory computer science courses to manage their students’ learning and success. Sometimes introductory college classes are notoriously called “weed-out” courses, which students who fall behind, are discouraged from continuing the career path. Students with different educational backgrounds may be unnecessarily defeated in these courses. In this work, we identify what class data can be collected and supplied to data analytic tooling to generate insights into monitoring the students’ progress. We first investigate a variety of machine learning tools and techniques on a class dataset. Then, we present work-in-progress designs of a visual analytic workflow. Through the interaction with the visual analytic tool, instructors of the introductory computer science course gather insights into the class, for example, “Which part of the programming assignment is causing students to have the most software bugs?,” “Which exam questions best test the understanding of runtime analysis?,” “What type of student activity results in fascinating over 50% of the class to participate and understand the material?”