Abstract Software engineering education widely adopts Agile and project-based learning approaches. However, questions remain about how student roles, perceptions, and the use of Large Language Models (LLMs) relate to learning outcomes. In this exploratory study, we investigate how students engage in a semester-long project-based agile course by examining three core elements: their software development roles and Scrum roles, their expectations and actual use of LLMs, and their perceptions of teamwork and learning practices. We focus on differences between below-average and above-average performing students, thereby linking project outcomes to effort, perceptions, and role-specific engagement. Our analyses show that student effort is strongly associated with performance, and that the distribution of typical software development roles relates to meaningful differences in both engagement and outcomes. The findings show that front-end developers tend to exhibit lower engagement and performance, indicating a need for role-specific support. Although students were expected to use LLMs, actual usage was limited, especially in complex tasks like testing and deployment, revealing a clear expectation–use gap rather than allowing conclusions about its underlying causes. Higher-performing students reported more positive perceptions of team collaboration and Scrum practices, suggesting a link between engagement, perception, and outcomes. Together, these findings contribute to understanding how roles, perceptions, and LLM expectations interact with student performance in agile, project-based settings.

Abstract Effort estimation is an important activity in agile software development. The goal of the presented study was to determine the influence of individual competence on software development effort estimation. In particular, we measured both the accuracy of effort estimation and the duration of the estimation process itself, both for three different estimation methods. The subjects of our study were teams of students of a graduate‐level software engineering course at the University of Ljubljana, Faculty of Computer and Information Science. Based on the grades that individual students attained in their undergraduate study, we classified each team as “high‐competence” or “low‐competence” and additionally as “heterogeneous” or “homogeneous” (the criterion here being the variance of the members' average grades). We found out that there was no significant difference in effort estimation accuracy neither between high‐competence and low‐competence teams nor between heterogeneous and homogeneous teams, regardless of which estimation method was used. However, high‐competence teams spent significantly less time on effort estimation than low‐competence ones. Likewise, for two of the employed estimation methods, heterogeneous teams completed effort estimation in a significantly shorter time than homogeneous teams. These results might benefit both academic and professional community.

Published studies on agile effort estimation predominantly focus on comparisons of the accuracy of different estimation methods, while efficiency comparisons, i.e. how much time the estimation methods consume was not in the forefront. However, for practical use in software development, the time required can be a very important cost factor for enterprises, especially when the accuracy of different agile effort estimations is similar. In this study, we thus try to advance the current standard accuracy comparison between methods by introducing efficiency, i.e. time it takes to use a method as an additional dimension of comparison. We conduct this comparison between three agile effort estimation methods that were not yet compared in the literature, namely, Planning Poker, Bucket System and Affinity Estimation. For the comparison, we used eight student teams with 29 students who had to use all the effort estimation methods during the course where they had to finish a programming project in 3 weeks. The results indicate that after the students get used to using the different methods the accuracy between them is not statistically significantly different, however, the efficiency is. On average, Bucket System and Affinity Estimation methods take half as much time as Planning Poker.

We describe a heuristic search-based method for interleaved HTTP (Web) session reconstruction building upon first order Markov models. An interleaved session is generated by a user who is concurrently browsing the same web site in two or more web sessions (browser tabs or windows). In order to assure data quality for subsequent phases in analyzing user's browsing behavior, such sessions need to be separated in advance. We propose a separating process based on best-first search and trained first order Markov chains. We develop a testing method based on various measures of reconstructed sessions similarity to original ones. We evaluate the developed method on two real world click stream data sources: a web shop and a university student records information system. Preliminary results show that the proposed method performs well.