Abstract
As programming education becomes important in the future, we need to solve problems such as the shortage of teachers and the increasing burden on work activities. In this study, we collect error messages from students in a programming lecture for private liberal arts college students. The characteristic of the error messages is analyzed using fuzzy graphs. The purpose of the analysis is to obtain knowledge for use in teaching learners from the results and to reduce the burden on teachers. In the analysis, the relationship between each error message is visualized using fuzzy graphs, and the characteristics of the students are extracted from the state of the clusters according to the degree of relationship. Regardless of whether the students are good or bad at English, we clarify clusters of students with many function-related errors and input errors, and examine the teaching methods for students. Based on the characteristics of the clusters and the changes in the clusters, it is possible to analyze the results as follows: "Students who simply make many errors and are likely to be inat-tentive should be given more attention and guidance to input their answers correctly". By using fuzzy graph analysis, efficient programming education and reduction of teachers' workload can be expected.
https://ror.org/01dq60k83 
