Abstract

After the popularity of Learning Management Systems, Data Mining and Learning Analytics have become emerging topics. Learning Management Systems such as Moodle, provide big amount of data to be used in analyzing students online behavior. This paper represents a method for early detection of drop outs from a Bachelor degree course using data mining methods. Data is collected through Moodle logs. For early detection, event logs till the first exam is taken into consideration. Decision Tree (DT) and Bayesian Network (BN) algorithms are used for the prediction. In the end it is shown that DT algorithm gives a higher over-all accuracy but BN is better for discovering fail cases as it has higher specificity.

References

[1] S. Dumais, R. Jeffries, D. M. Russell, D. Tang, and J. Teevan, “Understanding UserBehavior Through Log Data and Analysis,” in Ways of Knowing in HCI, New York, NY:Springer New York, 2014, pp. 349–372.
[2] G. Akçapınar, “Profiling Students’ Approaches to Learning through Moodle Logs,”Proceedings of Multidisciplinary Academic Conference on Education, Teaching and Elearning in Prague 2015, Czech Republic (MAC-ETeL 2015), no. December, p. 7, 2015.
[3] A. Konstantinidis and C. Grafton, “Using Excel Macros to Analyse Moodle Logs,” UKResearch.Moodle.Net, no. September, pp. 4–6, 2013.
[4] Á. Figueira and Álvaro, “Mining Moodle Logs for Grade Prediction,” in Proceedings ofthe 5th International Conference on Technological Ecosystems for EnhancingMulticulturality - TEEM 2017, 2017, pp. 1–8.
[5] Á. Figueira, “Mining Moodle Logs for Grade Prediction: A Methodology Walk-through,”in Proceedings of the 5th International Conference on Technological Ecosystems forEnhancing Multiculturality, 2017, pp. 44:1-44:8.
[6] T. Käser, N. R. Hallinen, and D. L. Schwartz, “Modeling exploration strategies to predictstudent performance within a learning environment and beyond,” in Proceedings of theSeventh International Learning Analytics & Knowledge Conference on - LAK ’17, 2017.
[7] M. Cocea and S. Weibelzahl, “Log file analysis for disengagement detection in e-Learningenvironments,” User Modeling and User-Adapted Interaction, 2009.
[8] N. Ademi and S. Loshkovska, “Exploratory Analysis of Student Activities and SuccessBased On Moodle Log Data,” in 16th International Conference on Informatics andInformation Technologies, 2019.(in press)
[9] K. Almohammadi, H. Hagras, D. Alghazzawi, and G. Aldabbagh, “A Survey of ArtificialIntelligence Techniques Employed for Adaptive Educational Systems Within E-Learning,”vol. 7, no. 1, pp. 47–64, 2017.
[10] T. Kaser, S. Klingler, A. G. Schwing, and M. Gross, “Dynamic Bayesian Networks for Student Modeling,” IEEE Transactions on Learning Technologies, vol. 10, no. 4, pp. 450–462, Oct. 2017.
[11] “RStudio - RStudio.”
[Online]. Available: https://rstudio.com/.
[Accessed: 13-Oct-2019].
[12] G. Grothendieck, “CRAN - Package sqldf.”
[Online]. Available: https://cran.rproject.org/web/packages/sqldf/index.html.
[Accessed: 12-Jun-2019].
[13] M. Hall, E. Frank, G. Holmes, B. Pfahringer, P. Reutemann, and W. I. H., “The WEKAData Mining Software: An Update,” SIGKDD Explorations, vol. 11, no. 1, 2009.
[14] M. Sokolova and G. Lapalme, “A systematic analysis of performance measures forclassification tasks,” Information Processing and Management, vol. 45, pp. 427–437,2009.