High dropout rates have been a major problem affecting the development of Massive Open Online Courses (MOOCs). Student dropout prediction can help teachers identify students who are tending to fail and provide extra help in a timely manner, helping to improve the effectiveness of online learning. In recent years, the use of convolutional neural networks for dropout prediction has yielded good results. However, traditional convolutional neural networks use automatic feature extraction, which does not consider the importance of the learner's behavior features and the effect of the time series of behavior on dropout, so it is difficult to guarantee the final prediction effect. To solve this problem, this article proposes a convolutional neural network model FWTS-CNN that integrates feature weighting and behavioral time series. It extracts continuous behavioral features from the learner's log of learning activities, filters key features and ranks them by importance based on the decision tree, then weights the continuous behavioral features based on importance, and finally builds a convolutional neural network model based on behavioral time series and weighted features. Experiments on the KDD Cup 2015 dataset show that the FWTS-CNN dropout prediction model has a high accuracy, which can reach more than 87%, an improvement of about 2% over using the CNN algorithm alone. The FWTS-CNN model integrates the effects of behavioral features and behavior time on dropout, effectively improving the accuracy of dropout prediction.

Learning programming challenges students who encounter difficulties such as resolving syntax and format errors. These challenges require students to invoke resilience to overcome programming problems and keep trying. In response, this study developed the programming resilience scale for university students (PRSUS). The snowball-sampling method was used to collect the data of science and engineering undergraduates, and the participants were divided into two groups (n1 = 316, n2 = 358) for an exploratory and a confirmatory factor analysis, respectively. For the PRSUS, 3 items were retained for each of 4 dimensions, namely: persistence, perceived value, difficulty cognition, and incremental belief. The reliability and validity indexes indicate the scale is reliable for measuring programming resilience. Moreover, the results showed that most undergraduates had a high level of programming resilience, and they could judge their level correctly in comparison to their peers. The study also found that, although no significant difference was found among the grade or age groups, the programming resilience of the male students was significantly better than that of the female students. The PRSUS is expected to help researchers and teachers to identify students who have difficulties learning programming and to provide early interventions to students. Implications for practice or policy: Students can use the PRSUS to test if they have a high level of programming resilience and improve their programming resilience in a targeted way. Educators can use the PRSUS to identify students who find it difficult to finish programming learning or to judge whether their teaching can make students more willing to overcome programming difficulties.

Large Language Models (LLMs) show great potential in programming learning. However, existing studies mainly focus on technical implementations and lacks a systematic analysis of the application of LLMs in programming learning from an educational perspective. This study conducts a systematic literature review and bibliometric analysis based on 75 high-quality papers, using a six-dimensional framework (roles, technology, learners, environment, effectiveness, challenges) to examine the current state and agenda of LLM applications. The results indicate that the application of LLMs has evolved from model validation in 2022 to teaching applications in 2023, and is expected to be deeply integrated into the education system by 2024-2025, reflecting a shift from tools to teaching agents. In programming learning, LLMs primarily take on roles in resource generation, task solving, and feedback provision. In terms of technology usage, OpenAI's series of models dominate, with Python being the main programming language environment, and research subjects focusing on beginner programmers and university students. Empirical studies show that LLMs can effectively enhance learners' cognitive outcomes and non-cognitive performance, but they can also lead to over-reliance on tools, academic integrity risks, and ethical challenges. Future research should establish an education theory-driven design framework for LLMs, conduct studies on generative AI literacy and ethical norms, and provide theoretical and practical guidance for programming learning.