CodeTutor: personalized programming learning through automated feedback and clustering

First-year Computer Science students often struggle to learn Java programming, and this condition can have a negative impact on their academic progress. Recently, an online tutoring system has been developed to help students learn Java programming. It allows students to execute and check Java code, making the learning process easier. Although the use of this tool has led to improvements in learning, it has been observed that the system has some weaknesses that need to be improved. The gap consists in the lack of specific and personalised feedback for the students. Therefore, a specific software has been developed, with an innovative approach based on machine learning using auto-clustering and neural networks. The new online tutoring system offers a better personalisation thanks to a graph model tailored to the study plan based on individual performance, by optimising learning effectiveness through video tutorials and Java code evaluations. This system aims not only to increase the academic success of the students, but also to provide instructors with a more precise assessment tool.