Utilizing student activity patterns to predict performance
Computer programming courses require a lot of practice. Nevertheless, the difficulties found when trying to install the computer programming environment used in the university courses prevent many students from practicing at home. Hence, they can only practice during the computer laboratory lessons, or when the laboratories are available for the students. This limitation negatively affects the learning process. In order to reduce this obstacle, we have replaced the conventional IDE used in our courses by a web-based IDE called Codeboard. Among other advantages, this environment has the ability to collect statistics about the number of accesses to the developed projects, the number of compilations and runs of the corresponding programs, and the number of global accesses per day. Thus, it enables teachers to monitor the activities of the students both in lectures and at home. After using this tool for a semester, in this paper we report the statistics that are collected when using Codeboard, the limitations we have found in this tool, and how they could be easily overcome in order to perform a better analysis. We also report on the results of this preliminary experience by comparing the results of three consecutive courses (taught by the same instructors and using the same materials) in order to evaluate the influence of Codeboard on the learning process.