Predicting Student Exam’s Scores by Analyzing Social Network Data

Fire, Michael; Katz, Gilad; Elovici, Yuval; Shapira, Bracha; Rokach, Lior

doi:10.1007/978-3-642-35236-2_59

Cited by 43 publications

(35 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Membership in some subcommunities defined by the MMSB are significantly predictive of dropout, while others are not; the number of subcommunities that are significant predictors varies between two and four across their three-MOOC sample (the authors consider up to 20 subcommunities per course). Other work has identified social networks as effective predictors of student performance in traditional academic courses (Fire et al, 2012;Gašević et al, 2013).…”

Section: Social Modelsmentioning

confidence: 99%

Student success prediction in MOOCs

Gardner

Brooks

2018

User Model User-Adap Inter

153

View full text Add to dashboard Cite

Predictive models of student success in Massive Open Online Courses (MOOCs) are a critical component of effective content personalization and adaptive interventions. In this article we review the state of the art in predictive models of student success in MOOCs and present a categorization of MOOC research according to the predictors (features), prediction (outcomes), and underlying theoretical model. We critically survey work across each category, providing data on the raw data source, feature engineering, statistical model, evaluation method, prediction architecture, and other aspects of these experiments. Such a review is particularly useful given the rapid expansion of predictive modeling research in MOOCs since the emergence of major MOOC platforms in 2012. This survey reveals several key methodological gaps, which include extensive filtering of experimental subpopulations, ineffective student model evaluation, and the use of experimental data which would be unavailable for real-world student success prediction and intervention, which is the ultimate goal of such models. Finally, we highlight opportunities for future research, which include temporal modeling, research bridging predictive and explanatory student models, work which contributes to learning theory, and evaluating long-term learner success in MOOCs.

show abstract

Section: Social Modelsmentioning

confidence: 99%

Student success prediction in MOOCs

Gardner

Brooks

2018

User Model User-Adap Inter

153

View full text Add to dashboard Cite

show abstract

“…Another cluster of recent research has examined the use of social interaction and participation data generated during the course, with mixed results. Fire et al analyzed social network graphs and found the strongest predictor from the graph to be a students "best" or nearest friend [13]. El-badrawy et al found online forum and discussion activity to be of low importance, while metrics related to accessing course material were of high importance [11], and Luo et al [21] found comments by students on the course and labs to be a strong predictor of course grades.…”

Section: Machine Learning Approachesmentioning

confidence: 99%

Evaluating Neural Networks as a Method for Identifying Students in Need of Assistance

Castro-Wunsch

Ahadi

Petersen

2017

Proceedings of the 2017 ACM SIGCSE Technical Symposium on Computer Science Education

View full text Add to dashboard Cite

Course instructors need to be able to students in need of assistance as early in the course as possible. Recent work has suggested that machine learning approaches applied to snapshots of small programming exercises may be an effective solution to this problem. However, these results have been obtained using data from a single institution, and prior work using features extracted from student code has been highly sensitive to differences in context. This work provides two contributions: first, a partial reproduction of previously published results, but in a different context, and second, an exploration of the efficacy of neural networks in solving this problem. Our findings confirm the importance of two features (the number of steps required to solve a problem and the correctness of key problems), indicate that machine learning techniques are relatively stable across contexts (both across terms in a single course and across courses), and suggest that neural network based approaches are as effective as the best Bayesian and decision tree methods. Furthermore, neural networks can be tuned to be reliably pessimistic, so they may serve a complementary role in solving the problem of identifying students who need assistance.

show abstract

“…Typically, the goal of education in programming is the acquisition of knowledge of programming, as well as, to some extent, the development of programming skills. Investigations concerning education have been conducted along the lines of general academic performance (Bergin & Reilly, 2005;Butcher & Muth, 1985;Byrne & Lyons, 2001), e↵ect of cognitive, behavioral, and attitudinal factors on learning outcomes (Fincher et al, 2005), knowledge of other programming languages prior to starting education (Hagan & Markham, 2000;Holden & Weeden, 2004), gender (Goold & Rimmer, 2000;Pioro, 2006), academic background (Pioro, 2006), and ability to trace (and explain) code (Lister, Fidge, & Teague, 2009), as well as using the exam scores of friends to predict the score of each student (Fire, Katz, Elovici, Shapira, & Rokach, 2012). Generally, the correlation between job performance and academic grades appear to be modest.…”

Section: Research On Programmers and Their Performancementioning

confidence: 99%

Inferring Skill from Tests of Programming Performance: Combining Time and Quality

Bergersen

Hannay

Sjøberg

et al. 2011

2011 International Symposium on Empirical Software Engineering and Measurement

View full text Add to dashboard Cite

Skilled developers are important to the software industry. In empirical studies in software engineering, knowing the skill level of the participants is also important for correct interpretation of results. The current practice in industry and research for assessing programming skills is mostly to use proxy variables of skill, such as education, experience, and multiple-choice knowledge tests. There is as yet no valid and e cient way to measure programming skill. Consequently, this thesis aimed to construct a valid instrument for measuring programming skill, where skill is inferred from performance on programming tasks. The Rasch measurement model was used to construct the instrument. Sixty-five professional developers from eight countries participated in validating the instrument, solving 19 Java programming tasks over two days. The validity of the instrument was theoretically investigated through commercial and research-based tests. Programming skill, as measured by the instrument, was also investigated in terms of experience and other background variables. The instrument was found to have desirable psychometric properties, and the overall results appear well aligned with theoretical expectations. This work has shown that acceptable measures of programming skill may be obtained with less than one day of testing. Further work should be directed at reducing the time needed to measure programming skill without a↵ecting the validity of the instrument. The results of the research have already been transferred to the industry through a commercial prototype. iii First and foremost, I am profoundly thankful to my principal supervisor Dag Sjøberg. He has provided me with unwavering support during my research for this thesis, often at odd hours and in strange places. I feel privileged to have had the opportunity to occupy so much of Dag's time over the years. The making of this thesis has been a truly rewarding experience, both professionally and personally, because of him. I am also grateful to my second supervisor, Tore Dybå, for interesting discussions and for inspiring me to look to other research fields for solutions to my research problem. Other researchers have also been important at various stages. I thank Erik Arisholm for his guidance and flexibility during the early years of my work. A special thanks goes to Jo Hannay and Magne Jørgensen, who are highly inspiring researchers. I am also grateful to Jan-Eric Gustafsson for his interest in my work. This thesis would have been impossible without financial support. Simula Research Laboratory gave me the opportunity to carry out the research and Simula Innovation and Simula School of Research and Education provided support as well. The FORNY program at the Norwegian Research Council funded parts of my work, while the University of Oslo allowed me to finish the thesis at a pace that was rewarding for me. This thesis would similarly have been impossible without technical support. Early on, Steinar Haugen shared his passion for programming with me and has since become a...

show abstract

Predicting Student Exam’s Scores by Analyzing Social Network Data

Cited by 43 publications

References 22 publications

Student success prediction in MOOCs

Student success prediction in MOOCs

Evaluating Neural Networks as a Method for Identifying Students in Need of Assistance

Inferring Skill from Tests of Programming Performance: Combining Time and Quality

Contact Info

Product

Resources

About