A methodology for visually representing student C++ programming proficiency

The grading time required to evaluate student programs and the quality of programming performance feedback to students could be improved if less subjective metrics could be applied using an automated grading tool to assist the grader. Before such a tool can become reality. additional research must be performed to define the metrics required to quantify a student's programming performance clearly. The purpose of this thesis is to develop and test a methodology to quantify a student's performance on programming assignments and represent that performance visually in a manner that is meaningful to the student. The research methodology used is to collect the student programs; extract the metrics data; filter the data; identify a set of metrics; group the selected metrics into complexity, quality, and style metric groups; establish bounds on these metrics; develop equations to represent program quality, complexity, and style; and develop a method to visualize student programming performance at both the module and program levels. The proposed methodology is successfully applied to all three CS 1463 programming assignments. Student programming assignment grades are compared with the assigned color consensus ratings computed by the proposed methodology, demonstrating that the color consensus program can distinguish between good and poor student programs.