| dc.description.abstract | In recent programming teaching, the online programming system has become a common method, allowing students to practice coding regardless of time and place. In addition, with the increasing demand for distance courses in response to the impact of Covid-19, students need to independently adjust their learning without face-to-face guidance from teachers. In past studies, it has been found that behavioral analysis is better than test scores to predict the ability of self-regulated learning, and programming behavioral analysis can also help teachers better understand students′ learning status and coding process. In addition, the application of the online programming system is divided into program evaluation in education, online compiler, etc. In the past research, most of them only analyzed the behavior of a single application, and most of them only analyzed the characteristics of the number of answers, the number of errors, and the time, and the analysis results tend to ignore the program content.
In this study, students use the programming behavior in the online programming system to define various states in practice and evaluation, generate the programming behavior sequence and convert it into a probability transition matrix, through the K-Means++ clustering algorithm separated 9 coding modes. In order to further study the coding modes, the correlation analysis was carried out between the probability of students staying in each coding mode and the scores of the programming test. Then, students with 3 different coding patterns (adept, struggle, and abandonment group) were distinguished by the students′ staying probability in each coding mode. In order to understand the characteristics of students of each coding pattern, we conducted an ANOVA analysis of the students of each coding pattern in the programming exam scores, self-regulated learning, and the probability of staying in each coding mode. Finally, according to the characteristics of students with different coding patterns, give corresponding feedback, and recommend the corresponding practice questions for students who are extremely troubled by logic and syntax and have a high probability of staying in the mode.
The study results show that the probability of staying in each coding mode is related to the programming exam scores and self-regulation learning to varying degrees. Although there is no significant difference in the programming exam scores of students with different coding patterns, the probability of staying in each coding mode is significant differences in different degrees in the ability to rehearsal, organization, and control beliefs. In terms of the effect of the intervention, the students in the experimental group were more familiar with the syntax of the program than before the intervention, and were significantly better than the control group in the performance of the programming exam. In addition, there was no significant difference in all aspects of self-regulated learning. | en_US |