Visual Analysis of Programming and Debugging Behavior for College Programming Courses

The behavior analysis of programming and debugging for college students is of great significance for teachers to optimize curriculum arrangement and improve students’ programming ability. Traditional tools for programming and debugging behavior analysis lack a collaborative visual analysis in terms of course types, topic categories and real-time changes of data relationships in memory, which makes it impossible to accurately describe students’ programming portraits and conduct students’ self-evaluation. According to the characteristics of programming and debugging behavior, we design a visual analysis system with multi-view collaborative interaction called MPDVAS. First, the spatial and temporal distribution of classes, courses, assignments and examinations in semesters and submission locations is shown through multi-dimensions radial heatmap-radar chart. Then a hierarchical bubble chart visualization method based on topic model with multi-platform online course data is proposed, which students are clustered into subgroups for different program ming behavior characteristics according to different portraits. Furthermore, we expand the sankey diagram for quantitative analysis and interactive reasoning of courses, grades and programming behaviors. Finally, a novel layout is proposed to give topics and courses recommendation by combining symmetrical stacked histogram and multi-dimensional time series charts with real-time evaluation process of code debugging and automatic comparison of program results. Through case analysis with real programming data of 313 students, feedbacks from 2 managers, 2 teachers and 20 students are collected for variance analysis. The p-value is 0.008 less than the significance level 0.05, which verified the effectiveness and practicability of MPDVAS.