A joint neural network model for combining heterogeneous user data sources: An example of at‐risk student prediction

Abstract

Information service providers often require evidence from multiple, heterogeneous information sources to better characterize users and offer personalized service. In many cases, statistic information (for example, users' profiles) and sequentially dynamic information (for example, logs of interaction with information systems) are two prominent sources that can be combined to achieve optimized results. Previous attempts in combining these two sources mainly exploited models designed for either static or sequential information, but not both. This study aims to fill the gap by proposing a novel joint neural network model that can naturally fit both static and sequential user data. To evaluate the effectiveness of the proposed method, this study uses the problem of at‐risk student prediction as an example where both static data (personal profiles) and sequential data (event logs) are involved. A thorough evaluation was conducted on an open data set, with comparisons to a range of existing approaches including both static and sequential models. The results reveal superb performances of the proposed method. Implications of the findings on further research and applications of joint models are discussed.

An Interpretable Pipeline for Identifying At-Risk Students

“…This approach employed a neural network structure to keep track of the trends of students' attrition. Similarly, Qiao and Hu (2020) implemented a joint neural network model to analyze both user demographic information and learning behaviors for predicting at-risk students. The experimental results demonstrated the proposed model outperformed the commonly used baseline models in identifying at-risk students.…”

“…To demonstrate the validity of the proposed method, we used the OULAD (see https://analyse.kmi.open.ac.uk/open_dataset) dataset collected by Open University (Kuzilek et al., 2017). Multiple studies (Adnan et al., 2021; Aljohani et al., 2019; Hlosta et al., 2018, 2017; Qiao & Hu, 2020) have been conducted using this dataset as a baseline to verify the effectiveness of their methods in predicting student learning performance. In this study, we selected four courses purposely from the dataset, where two of them (i.e., AAA_2013J and AAA_2014J) were in the social science field and the other two (i.e., CCC_2014B and CCC_2014J) were in the STEM field.…”

Assignments as Influential Factor to Improve the Prediction of Student Performance in Online Courses

“…However, this study has important limitations, like the fact of analyzing only two courses of a total of seven available. The other studies that use assignment information, a 30% of works, use this factor together with the rest of sources sources [3,39,[41][42][43]. Thus, the real relevance of this factor in the final prediction cannot be analyzed.…”

“…Regarding the purpose of the different works, under the main task of predicting student performance, it can be found that the majority of studies pretend to predict whether the student will pass or fail a course [3,27,[30][31][32][37][38][39][40][41][42][43][44][45]. Other approaches focus on the dropout rate [26,29,32,33], while others follow an early prediction study [33,35,36,46].…”

“…With regard to the algorithms used, it can be appreciated that different supervised learning techniques have been used, mainly trees-based methods [28,30,31,33,36,40,44,46], although Bayesian methods have also been applied [38], as well as support vector machines [27], generative methods based on sequences [26,32] and neural networks or deep learning [3,34,35,39,[41][42][43]45].…”