PerformanceVis: Visual analytics of student performance data from an introductory chemistry course

Abstract

We present PerformanceVis, a visual analytics tool for analyzing student admission and course performance data and investigating homework and exam question design. Targeting a university-wide introductory chemistry course with nearly 1000 student enrollment, we consider the requirements and needs of students, instructors, and administrators in the design of PerformanceVis. We study the correlation between question items from assignments and exams, employ machine learning techniques for student grade prediction, and develop an interface for interactive exploration of student course performance data. PerformanceVis includes four main views (overall exam grade pathway, detailed exam grade pathway, detailed exam item analysis, and overall exam & homework analysis) which are dynamically linked together for user interaction and exploration. We demonstrate the effectiveness of PerformanceVis through case studies along with an ad-hoc expert evaluation. Finally, we conclude this work by pointing out future work in this direction of learning analytics research. Keywords: Student performance, Item analysis, Grade prediction, Learning analytics, Knowledge discovery