Figure 5 - uploaded by Wanli Xing
Content may be subject to copyright.
Source publication
Massive open online courses (MOOCs) show great potential to transform traditional education through the Internet. However, the high attrition rates in MOOCs have often been cited as a scale-efficacy tradeoff. Traditional educational approaches are usually unable to identify such large-scale number of at-risk students in danger of dropping out in ti...
Contexts in source publication
Context 1
... learning is a format of neural network that takes metadata as input and then processes the data through a number of layers to compute the output ( LeCun et al., 2015). While traditional neural network can only handle single hidden layer ( Figure 5, left), deep learning processes the input data through a large number of hidden layers in its structure ( Figure 5, right). Each layer is made of nodes, which is the place for computation to take place. ...
Context 2
... learning is a format of neural network that takes metadata as input and then processes the data through a number of layers to compute the output ( LeCun et al., 2015). While traditional neural network can only handle single hidden layer ( Figure 5, left), deep learning processes the input data through a large number of hidden layers in its structure ( Figure 5, right). Each layer is made of nodes, which is the place for computation to take place. ...
Context 3
... learning is a format of neural network that takes metadata as input and then processes the data through a number of layers to compute the output ( LeCun et al., 2015). While traditional neural network can only handle single hidden layer ( Figure 5, left), deep learning processes the input data through a large number of hidden layers in its structure ( Figure 5, right). Each layer is made of nodes, which is the place for computation to take place. A node com- bines input from the data with a set of weights to determine whether to amplify or dampen the input, which in turn assigns significance to the inputs. These input weight products are then summed and evaluated to decide to what extent the information propagates through the network to finally influence the classification. In a more holistic view, the hidden layer trains the unique set of features using the output of the previous layer. This process is known as non- linear transformation. The more hidden layers it has, the more complex and abstract the data will ...
Context 4
... learning is a format of neural network that takes metadata as input and then processes the data through a number of layers to compute the output ( LeCun et al., 2015). While traditional neural network can only handle single hidden layer ( Figure 5, left), deep learning processes the input data through a large number of hidden layers in its structure ( Figure 5, right). Each layer is made of nodes, which is the place for computation to take place. A node com- bines input from the data with a set of weights to determine whether to amplify or dampen the input, which in turn assigns significance to the inputs. These input weight products are then summed and evaluated to decide to what extent the information propagates through the network to finally influence the classification. In a more holistic view, the hidden layer trains the unique set of features using the output of the previous layer. This process is known as non- linear transformation. The more hidden layers it has, the more complex and abstract the data will ...
Similar publications
Die vorliegende Arbeit beschäftigt sich mit dem Thema des Einsatzes digitaler Medien in Volksschulen und stellt die Erhebung eines Ist-Standes in unterschiedlichen Bereichen dar. Sie gibt einen Überblick über derzeitige Maßnahmen zur digitalen Bildung, setzt sich-durch das Beschreiben unterschiedlicher Sichtweisen-kritisch mit dem Einsatz digitaler...
Textual data, as a key carrier of learning feedback, is continuously
produced by many students within course forums. The temporal nature of discussion requires students’ emotions and concerned aspects (e.g. teaching styles, learning activities, etc.) to be dynamically tracked for understanding learning requirements. To characterize dynamics of emot...
Learning Analytics ist die Möglichkeit aus gesammelten Daten durch entsprechende Analyseverfahren neue Erkenntnisse für die pädagogische Praxis zu erhalten. In diesem Beitrag soll eine kurze Einführung gegeben werden, sowie anhand von zwei konkreten Beispielen gezeigt werden wir man es für den Schulalltag nutzen kann.
Citations
... One of the most important areas of EDM is the prediction of learning behavior and the outcome of students (e.g., [19,20]). Tis can further be categorized into three forms of predictive problems ( [21]), including the prediction of students' performance (e.g., [19,20]), the prediction of students' dropout (e.g., [22,23]), and the prediction of students' grades or achievement in a course (e.g., [24]). ...
While modeling students’ learning behavior or preferences has been found to be a crucial indicator for their course achievement, very few studies have considered it in predicting the achievement of students in online courses. This study aims to model students’ online learning behavior and accordingly predict their course achievement. First, feature vectors are developed using their aggregated action logs during a course. Second, some of these feature vectors are quantified into three numeric values that are used to model students’ learning behavior, namely, accessing learning resources (content access), engaging with peers (engagement), and taking assessment tests (assessment). Both students’ feature vectors and behavior models constitute a comprehensive student’s learning behavioral pattern which is later used for the prediction of their course achievement. Lastly, using a multiple-criteria decision-making method (i.e., TOPSIS), the best classification methods were identified for courses with different sizes. Our findings revealed that the proposed generalizable approach could successfully predict students’ achievement in courses with different numbers of students and features, showing the stability of the approach. Decision tree and AdaBoost classification methods appeared to outperform other existing methods on different datasets. Moreover, our results provide evidence that it is feasible to predict students’ course achievement with high accuracy through modeling their learning behavior during online courses.
... Its predictive capability aligns with proven educational models that have been utilized to successfully inform teaching practices and support adaptive learning strategies [66]. The foundational accuracy of our model indicates a strong potential for real-world applicability, as seen in similar simulations that have influenced educational strategies even before empirical testing [84]. The simulator can support educators by delivering actionable insights that enhance personalized interventions, curriculum design, and evidence-based teaching practices. ...
Student simulation supports educators to improve teaching by interacting with virtual students. However, most existing approaches ignore the modulation effects of course materials because of two challenges: the lack of datasets with granularly annotated course materials, and the limitation of existing simulation models in processing extremely long textual data. To solve the challenges, we first run a 6-week education workshop from N = 60 students to collect fine-grained data using a custom built online education system, which logs students' learning behaviors as they interact with lecture materials over time. Second, we propose a transferable iterative reflection (TIR) module that augments both prompting-based and finetuning-based large language models (LLMs) for simulating learning behaviors. Our comprehensive experiments show that TIR enables the LLMs to perform more accurate student simulation than classical deep learning models, even with limited demonstration data. Our TIR approach better captures the granular dynamism of learning performance and inter-student correlations in classrooms, paving the way towards a ''digital twin'' for online education.
... A study by Wanli Xing et al. conducted research on how to improve dropout prediction models for MOOC by utilising DL algorithms [51]. The authors contended that high attrition rates result from traditional educational systems' frequent inability to promptly identify at-risk students. ...
Massive open online courses have revolutionised the learning environment, but their effectiveness is undermined by low completion rates. Traditional dropout prediction models in MOOCs often overlook complex factors like temporal dependencies and context-specific variables. These models are not adaptive enough to manage the dynamic nature of MOOC learning environments, resulting in inaccurate predictions and ineffective interventions. Accordingly, MOOCs dropout prediction models require more sophisticated artificial intelligence models that can address these limitations. Moreover, incorporating feature selection methods and explainable AI techniques can enhance the interpretability of these models, making them more actionable for educators and course designers. This paper provides a comprehensive review of various MOOCs dropout prediction methodologies, focusing on their strategies and research gaps. It highlights the growing MOOC environment and the potential for technology-driven gains in outcome accuracy. This review also discusses the use of advanced models based on machine learning, deep learning, and meta-heuristics approaches to improve course completion rates, optimise learning outcomes, and provide personalised educational experiences.
... Deep Learning (DL) [25] and Machine Learning (ML) [26] are the sub-domains of Artificial Intelligence (AI) which have been revolutionised for performing Learning Analytics (LA) [27] tasks. These LA tasks include feature subset selection, feature processing and designing prediction models for the academic performance prediction problems. ...
The increasing reliance on Massive Open Online Courses (MOOCs) has transformed the landscape of education, particularly during the COVID-19 pandemic, where e-learning became essential. However, the effectiveness of MOOCs in enhancing student academic performance and engagement remains a key challenge, compounded by high dropout rates and low retention. This study presents a systematic literature review (SLR) conducted over a five-year period (2019–2024) to identify factors affecting student academic performance and engagement prediction in MOOCs, utilizing Deep Learning (DL) methods. The review follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, systematically analyzing articles from five major academic databases: ScienceDirect, SpringerLink, Scopus, Taylor & Francis, and Wiley Online. A total of 70 articles were selected for in-depth analysis, focusing on key predictors of student performance and engagement, including demographic data, behavioral patterns, learning activities, and clickstream data. The review highlights the capabilities of DL techniques in predicting student outcomes, such as retention, dropout, and engagement, offering valuable insights for educators and policymakers aiming to improve MOOC-based learning environments. By conducting SLR using PRISMA model, we identified research findings and gaps by proposing a conceptual framework for developing future personalized and adaptive e-learning environment for the inclusive MOOC based deaf and blind learners. This paper concludes by discussing implications for future personalized and adaptive e-learning environments and the necessity of comprehensive teacher training programs to navigate these evolving educational technologies.
... Summative prediction uses all available course data at the semester's end to build a comprehensive and accurate model (Susnjak, 2024). While this method offers high accuracy, it cannot provide timely support or early interventions for struggling students (Xing & Du, 2019;Al-Shabandar et al., 2017). ...
... Early prediction models can identify at-risk students as soon as the first week of classes (Sraidi et al., 2022). However, these models may be less accurate due to incomplete data and may miss patterns that become evident later in the course (Riestra-González et al., 2021;Xing & Du, 2019). ...
Machine learning and data mining techniques hold promise in predicting at-risk students in online learning. This meta-analysis aimed to provide quantitative evidence to validate whether and to what extent machine learning techniques have been achieved in identifying online at-risk students. Meta-regressions examined the impacts of predictor data types, classical or deep learning approaches, and prediction stages on performance. A random-effects meta-analysis of 47 studies with 309 models showed good classification accuracy, higher in summative than early predictions. Deep learning models and diverse predictors can significantly enhance model performance. No publication bias was detected. Implications and recommendations for practice are discussed.
... The novel dropout prediction model is proposed by Xing et al [24]. The intervention personalization was examined for improving the effectiveness of the model in MOOCs. ...
... Panagiotakopoulos et al [23] proposed an early dropout prediction model and it achieve an accuracy of 91.00%. In comparison to this, Xing et al [24] and Chen et al [25] achieve an accuracy of 95.01% and 92.5% for dropout prediction. This approach is used to personalize and prioritize intervention for at-risk students in MOOCs by using individual dropout probabilities. ...
In the online education field, Massive open online courses (MOOCs) have become popular in recent years. Educational institutions and Universities provide a variety of specialized online courses that helps the students to adapt with various needs and learning preferences. Because of this, institutional repositories creates and preserve a lot of data about students' demographics, behavioral trends, and academic achievement every day. Moreover, a significant problem impeding their future advancement is the high dropout rate. For solving this problem, the dropout rate is predicted by proposing an Ensemble Deep Learning Network (EDLN) model depending on the behavior data characteristics of learners. The local features are extracted by using ResNet-50 and then a kernel strategy is used for building feature relations. After feature extraction, the high-dimensional vector features are sent to a Faster RCNN for obtaining the vector representation that incorporates time series data. Then an attention weight is obtained for each dimension by applying a static attention mechanism to the vector. Extensive experiments on a public data set have shown that the proposed model can achieve comparable results with other dropout prediction methods in terms of precision, recall, F1 score, and accuracy.
... Similarly, Predictive Modelling forecasts learning outcomes to identify students needing extra support and here AI is adept at forecasting future learning outcomes by analysing students' past performance [98][99][100]. This predictive capability is instrumental in the early identification of students who may require additional support, DRAFT-Preprint 13 thereby enabling timely intervention. ...
This paper investigates the transformative role of Artificial Intelligence (AI) in creating and enhancing ecological learning spaces. It explores how AI technologies are reshaping educational environments to become more adaptive, interconnected, and personalised. The study begins by tracing the evolution of Generative AI and its application in education, followed by an examination of the concept of ecological learning spaces. The chapter then examines some specific ways AI is impacting these spaces, focusing on three key aspects: dynamic adaptability, interconnectedness, and learner-centred design. It discusses how AI facilitates personalised learning experiences, enables collaborative networks, and supports holistic development and educational accessibility. The paper concludes by outlining the potential of AI in transforming educational experiences and suggesting future directions for research in this field.
... For instance, an adaptive predictive model was employed to develop an early warning system that offers timely feedback and intervention for students at risk of underperforming [38]. Additionally, a dropout prediction model was optimized for at-risk students in MOOCs by incorporating personalized interventions tailored to individual situations and preferences [39]. ...
he field of Learning Analytics (LA) has witnessed remarkable growth, with a growing emphasis on the utilization of data-driven insights to enhance educational practices. Learning Analytics, encompassing the acquisition, analysis, and interpretation of student data, holds immense promise in transforming education. This review paper synthesizes the key advancements in Learning Analytics, focusing on its definition, benefits, and various levels of learning analytics. A comprehensive literature review has been conducted to delve into existing platforms, LA levels, and technologies. It critically evaluates the significance of predictive Learning Analytics in identifying trends and patterns in educational data. Moreover, the review delves into the integration of Artificial Intelligence (AI) in LA, highlighting its multifaceted utility, from personalized recommendations to intelligent tutoring systems. Several case studies are examined to underscore the real-world applications of AI models in Learning Analytics. This paper offers insights into the advantages of AI-driven LA, such as early intervention and adaptive learning. Challenges and ethical considerations in AI-powered LA are also discussed. Furthermore, it shines a spotlight on the field of machine learning within Learning Analytics, emphasizing its role in automating data analysis and prediction, thus streamlining educational processes. This comprehensive review provides a foundational understanding of the evolving landscape of Learning Analytics, AI, and Machine Learning in education.
... Another aspect of LA includes the forecasting of student academic achievement [2,[10][11][12][13][14], the detection of patterns in system use and navigation, and the identification of students who may be prone to academic underperformance. Educational technology platforms, such as Massive Open Online Courses (MOOC), learning management systems (LMS), student information systems (SIS), and intelligent teaching systems (ITS), provide digital data that may be used to assess students' future behaviors. ...
Machine learning (ML) is an emerging field of study that utilizes data to enhance the learning process and optimize the learning environment. The primary goals of ML are to observe students’ activities and provide early predictions about their academic performance, with the aim of enhancing student retention. Furthermore, ML aims to provide personalized feedback and streamline the provision of support to pupils. A flipped classroom is an educational approach that integrates both physical and digital spaces, known as blended learning environments. Flipped classes often use learning management systems that provide access to recorded lectures and digital resources. This facilitates the collection of statistics on students’ interaction with these services. The present chapter used bibliometric analysis to examine the effect of ML in predicting students’ performance in flipped classes. Information was extracted from the Scopus database for the period of 2014–2024. The data were examined using the R statistical programming language and the Biblioshiny software. Through the use of this strategy, we are presented with possibilities to enhance our skills and expertise in the respective domain. The investigation reveals that ML systems provide automated data-driven formative feedback, which supports students’ self-regulation and enables instructors to identify areas and tactics for intervention and assistance.
... As we observed in our previous study, the prediction performance increases when more data (especially objective data) is considered in calculating the predictions. This is in line with the results of Xing and Du (2019), who found an increase in predictive performance when predicting dropouts in MOOCs. ...
Academic procrastination, i.e., the irrational delay of important academic
tasks, is a potentially harmful behaviour that is highly prevalent in higher
education (HE), especially distance education (DE). Given the wide range
of adverse outcomes, including mood, well-being and academic
performance, interventions might benefit many students. In this chapter,
we presented a framework that is aimed at reducing academic
procrastination. It consists of a conceptual framework based on a theoretical
foundation. We detailed how various concepts related to academic
procrastination are linked, the crucial role that self-directed learning (SDL)
can play, which intervention strategies have the highest potential and what
the associated predictive models should encompass. The framework’s
components are showcased through a use case3, namely, a Java
programming course. The chapter concludes with an outlook on future
research that is necessary to lay the foundation for implementing this
framework.