Predictive learning analytics for video-watching behavior in MOOCs

Shridharan, Madhumitha; Willingham, Ashley; Spencer, Jonathan; Yang, Tsung-Yen; Brinton, Christopher G.

doi:10.1109/ciss.2018.8362323

Cited by 10 publications

(7 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…for m = 1, ..., K p represents the loss between sample and label for p-th cluster, and meta (•) is the loss function of the weighting network. The nested structure in (9) specifies that the optimal Θ * is based on the loss of each cluster L meta,p at the optimal CFA prediction model w * , which in turn is dependent on Θ and L train . The gradient descent learning process of the classifier network (7) at time step t is formulated as:…”

Section: Clustering-guided Meta-learningmentioning

confidence: 99%

“…Contemporary eLearning platforms collect a substantial amount of data on student interactions. This brings novel opportunities to study the process of human learning, and in turn to improve user experience on eLearning platforms, e.g., through predictive learning analytics and content personalization [7]- [9]. Specifically, learning management systems employed at educational institutions are typically capable of collecting data on quiz/assessment responses, clickstream actions on user navigation through the platform, content access logs, social networking on discussion forums, and video watching behavior.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Click-Based Student Performance Prediction: A Clustering Guided Meta-Learning Approach

Chu¹,

Tenorio²,

Cruz³

et al. 2021

Preprint

View full text Add to dashboard Cite

We study the problem of predicting student knowledge acquisition in online courses from clickstream behavior. Motivated by the proliferation of eLearning lecture delivery, we specifically focus on student in-video activity in lectures videos, which consist of content and in-video quizzes. Our methodology for predicting in-video quiz performance is based on three key ideas we develop. First, we model students' clicking behavior via time-series learning architectures operating on raw event data, rather than defining hand-crafted features as in existing approaches that may lose important information embedded within the click sequences. Second, we develop a self-supervised clickstream pre-training to learn informative representations of clickstream events that can initialize the prediction model effectively. Third, we propose a clustering guided meta-learning-based training that optimizes the prediction model to exploit clusters of frequent patterns in student clickstream sequences. Through experiments on three real-world datasets, we demonstrate that our method obtains substantial improvements over two baseline models in predicting students' in-video quiz performance. Further, we validate the importance of the pre-training and meta-learning components of our framework through ablation studies. Finally, we show how our methodology reveals insights on video-watching behavior associated with knowledge acquisition for useful learning analytics.

show abstract

Section: Clustering-guided Meta-learningmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Click-Based Student Performance Prediction: A Clustering Guided Meta-Learning Approach

Chu¹,

Tenorio²,

Cruz³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…User behaviour prediction and analysis are also important for the system's decisionmaking. As discussed in the study [31], the way how a learner may interact is worthwhile to understand in order to provide fine-grained insights into what particular content may be improved for further modification or adaptation of the learning activities. But only investigating one course [28] is inadequate for training and validating the model proposed in [31].…”

Section: User Behaviour Predictionmentioning

confidence: 99%

“…As discussed in the study [31], the way how a learner may interact is worthwhile to understand in order to provide fine-grained insights into what particular content may be improved for further modification or adaptation of the learning activities. But only investigating one course [28] is inadequate for training and validating the model proposed in [31]. Another prior study [32] suggested that different watching patterns might represent different cognitive levels, where the users' next behaviours and future performance could be predicted by clicking interactions.…”

Section: User Behaviour Predictionmentioning

confidence: 99%

Towards the Readiness of Learning Analytics Data for Micro Learning

Lin

Sun

Shen

et al. 2019

Services Computing – SCC 2019

View full text Add to dashboard Cite

With the development of data mining and machine learning techniques, data-driven based technologyenhanced learning (TEL) has drawn wider attention. Researchers aim to use established or novel computational methods to solve educational problems in the 'big data' era. However, the readiness of data appears to be the bottleneck of the TEL development and very little research focuses on investigating the data scarcity and inappropriateness in the TEL research. This paper is investigating an emerging research topic in the TEL domain, namely micro learning. Micro learning consists of various technical themes that have been widely studied in the TEL research field. In this paper, we firstly propose a micro learning system, which includes recommendation, segmentation, annotation, and several learning-related prediction and analysis modules. For each module of the system, this paper reviews representative literature and discusses the data sources used in these studies to pinpoint their current problems and shortcomings, which might be debacles for more effective research outcomes. Accordingly, the data requirements and challenges for learning analytics in micro learning are also investigated. From a research contribution perspective, this paper serves as a basis to depict and understand the current status of the readiness of data sources for the research of micro learning.

show abstract

“…Additionally, some datasets lack the behavioral aspect of learners like the Academic Performance dataset [9]. In the Coursera platform, some researchers searched only in the discussion forums for analyzing the cognitive process [10], but others dealt with learners' clickstreams in videos for predicting the learners' future behavior [11]. Ultimately, this paper chooses the Open University Learning Analytics Dataset (OULAD), because the gathered data with anonymization [12] covers all the learners' individual differences including the demographic data, the summary of their daily activities when they interact with the VLE and course assessment outcome [13].…”

Section: Introductionmentioning

confidence: 99%

Online At-Risk Student Identification using RNN-GRU Joint Neural Networks

Chen

et al. 2020

Information

View full text Add to dashboard Cite

Although online learning platforms are gradually becoming commonplace in modern society, learners’ high dropout rates and serious academic performance require more attention within the virtual learning environment (VLE). This study aims to predict students’ performance in a specific course as it is continuously running, using the statistic personal biographical information and sequential behavior data with VLE. To achieve this goal, a novel recurrent neural network (RNN)-gated recurrent unit (GRU) joint neural network is proposed to fit both static and sequential data, where the data completion mechanism is also adopted to fill the missing stream data. To incorporate the sequential relationship of learning data, three kinds of time-series deep neural network algorithms: simple RNN, GRU, and LSTM are first taken into consideration as baseline models. Their performances are compared in identifying at-risk students. Experimental results on Open University Learning Analytics Dataset (OULAD) show that simple methods like GRU and simple RNN have better results than the relatively complex LSTM model. The results also reveal that different models have different peak performance time, which results in the proposed joint model that achieves over 80% prediction accuracy of at-risk students at the end of the semester.

show abstract

Predictive learning analytics for video-watching behavior in MOOCs

Cited by 10 publications

References 11 publications

Click-Based Student Performance Prediction: A Clustering Guided Meta-Learning Approach

Click-Based Student Performance Prediction: A Clustering Guided Meta-Learning Approach

Towards the Readiness of Learning Analytics Data for Micro Learning

Online At-Risk Student Identification using RNN-GRU Joint Neural Networks

Contact Info

Product

Resources

About