During the past two decades, higher education institutions have been experiencing challenges in transforming the traditional way of in-class teaching into blended learning formats with the support of e-learning technologies that make possible the collection and storing of considerable amounts of data on students. These data have considerable potential to bring digital technologies in education to a new level of personalized learning and data-driven management of the educational process. However, the way data are collected and stored in a typical university makes it difficult to achieve the mentioned goals, with limited examples of data being used for the purposes of learning analytics. In this work, based on the analysis of existing information systems and databases at Siberian Federal University, we propose principles of design for a university database architecture that allow for the development and implementation of a data-driven management approach. We consider various levels of detail of education data, describe the database organization and structure, and provide examples of learning analytics tools that can benefit from the proposed approach. Furthermore, we discuss various aspects of its implementation and associated questions.
Student retention is a significant challenge for higher education institutions (HEIs). The fact that a considerable number of dropouts from universities are primarily due to academic underperformance motivates universities to develop learning analytics tools based on models for predicting learning success. However, the scalability of such models is limited since students’ academic performance and engagement, as well as the factors influencing them, are largely determined by the educational environment. The article proposes a hybrid approach to forecasting success in completing an academic semester, which involves creating a set of predictive models. Some of the models use historical student data, while others are intended to refine the forecast using current data on student performance and engagement, which are regularly extracted from available sources. Based on this approach, we developed an ensemble of machine learning models and the Markov-process model that simultaneously address the tasks of forecasting success in mastering a course and success in completing a semester. The models utilize digital footprint data, digital educational history, and digital personality portraits of students extracted from the databases of Siberian Federal University, and the resulting ensemble demonstrates a high quality of the forecast. The proposed approach can be utilized by other HEIs as a framework for creating mutually complementary forecasting models based on different types of accessible educational data.
1 September 2019 in Journal of Siberian Federal University. Humanities & Social Sciences
The problem of individualisation in education in terms of e-learning remains relevant despite a lot of research done in the area. The present article is intended to review existing education practice individualisation in e-learning and to determine the prospects of development of adaptive e-learning courses that provide the student with an individual educational environment content which focuses on the individual characteristics of students. The authors arrive at the conclusion that building an individual educational trajectory in the adaptive e-learning course has a significant advantage in organising education in the electronic environment for both the student and the teacher. The main feature of these courses becomes the selection of educational content that best matches the individual abilities, needs and needs of the student in an electronic environment. The results of the research are universally applicable and can serve as a basis for organising adaptive web-based teaching not only for mathematics, but also for other disciplines in institutions of secondary, higher and further education
The use of learning success prediction models is increasingly becoming a part of practice in educational institutions. While recent studies have primarily focused on the development of predictive models, the issue of their temporal stability remains underrepresented in the literature. This issue is critical as model drift can significantly reduce the effectiveness of Learning Analytics applications in real-world educational contexts. This study aims to identify effective approaches for assessing the degradation of predictive models in Learning Analytics and to explore retraining strategies to address model drift. We assess model drift in deployed academic success prediction models using statistical analysis, machine learning, and Explainable Artificial Intelligence. The findings indicate that students’ Digital Profile data are relatively stable, and models trained on these data exhibit minimal model drift, which can be effectively mitigated through regular retraining on more recent data. In contrast, Digital Footprint data from the LMS show moderate levels of data drift, and the models trained on them significantly degrade over time. The most effective strategy for mitigating model degradation involved training a more conservative model and excluding features that exhibited SHAP loss drift. However, this approach did not yield substantial improvements in model performance.
1 September 2022 in Perpsectives of science and education
Introduction. The recent active development of digital educational infrastructure allows educational institutions to collect and store large amounts of data related to the learning process. In this regard, the task of describing, structuring and combining various data about the student into a single digital profile, seems relevant as it will make possible applying an integrated approach to solving the problems of data-driven management of the educational process. Materials and methods. A comparative analysis of the literature to analyze the existing terminology in the field of digital transformation of education and methods for modeling objects and processes in education, as well as to identify the components of the digital student's profile has been carried out. To assess the intensity of activity performed by students in the digital learning environment (DLE) of the university, we have conducted the survey of the 172 students of Siberian Federal University majoring in several areas of training: Computer Science and Computer Engineering, Applied Computer Science, Information Security, and Metallurgy. Results. We proposed the so-called student digitalization pyramid, which includes the following hierarchy levels: digital footprint, digital profile, digital model, digital twin. We described the structure of the digital profile of a student which includes two components – a digital personality portrait and a digital educational history and mentioned possible links between them. We introduced a classification of its components and identified the sources of obtaining these components from the university EIEE. We introduced and described in detail the concept of digital educational history which is a multidimensional structured dynamically changing data on the educational activities of the student and his current learning outcomes. We indicated the problem of completeness of digital educational history data, the presence of which is confirmed by the results of the survey. Three most frequently used sources of educational materials in addition to the university e-learning courses were identified as video hosting (used by 85.5% of respondents), educational interaction in small groups (62.2%) and learning forums (45.3%). At the same time, obtaining educational history data from these sources is a serious problem which is difficult to overcome. Conclusions. In the future, digital educational history data will become the basis for modeling a student (their learning process) and developing their digital twin. Nevertheless, there are a number of problems which concern the collection and verification of the related data, the ethics of their use and applicability for solving learning management problems. We propose possible solutions to these problems
Student retention prediction is one of the most important problems of learning analytics. In the global scope research on the topic for higher education is rather extensive, there are cases of successful implementation of education support services in universities. The literature analysis shows of the growing interest in this problem in the Russian scientific and pedagogical community. At the same time, the specifics of Russian education does not allow direct transfer of foreign experience into the domestic educational system.The study reveals that a significant contribution to predicting student retention can be made by models for predicting academic performance in educational courses of the curriculum. The authors propose a structural model of a system for predicting academic performance, which includes a universal model based on generalized indicators of the digital footprint, a course-based model that takes into account the specifics of learning in a particular discipline, and a model based on the student’s educational profile.In the empirical study we trained 5 models for early prediction of interim assessment grades based on the universal indicators of the LMS Moodle student digital footprint. The most accurate model, especially in the first half of the semester, turned out to be ensemble-averaging models of logistic regression, random forest and gradient boosting. It was found that universal models are effective for detection of at-risk students in the discipline, the directions for further improvement of the universal model of performance prediction were determined and conditions for scaling the proposed approach to create a prognostic system for student retention to other educational institutions were formulated.