Toward Human–AI Collaboration: A Recommender System to Support CS1 Instructors to Select Problems for Assignments and Exams

Learning Analytics in Introductory Programming Courses: A Showcase from the Federal University of Amazonas

“…In fact, by combining the number of attempts and correctness with highly granular variables, our predictive models achieved an F1-Score of approximately 80% (Accuracy ≈ 81%, Recall ≈ 82%, Precision ≈ 79%) in predicting student performance using only data from the first 2 or 3 weeks of class (Pereira et al, 2021). Furthermore, the predictive models become increasingly accurate as more data is accumulated from subsequent weeks (Pereira et al, 2022). To illustrate, by the sixth week, our predictive model had achieved an F1-Score of approximately 90%.…”

“…These students may feel bored or even frustrated. One potential solution would be to recommend more challenging programming activities to keep these students engaged in the course (Pereira et al, 2022).…”

“…This release covers data from 107 CS1 classes, 4,511 students, 14,947 homework exercises, 1,908 exams, 375,173 student codes, and 3,860,565 code tests and submissions. These data have enabled researchers in computer science education and LA to conduct research on educational online judges (Carvalho et al, 2016), code plagiarism detection (Oliveira et al, 2021), continuous student authentication (Lavareda et al, 2020), prediction of learning and academic performance in programming (Pereira et al, 2017;Fonseca et al, 2019;Pereira et al, 2019;Pereira, Souza, et al, 2020), interpretation of predictive models (Pereira et al, 2021), recommendation systems (Pereira et al, 2022), gamification (Pessoa et al, 2019(Pessoa et al, , 2021Ribeiro et al, 2020;Pessoa, 2022), among others. In addition, RBIE v.31 -2023 these research efforts contribute to the formation of new masters and doctors at the IComp at the UFAM.…”