Learner classification based on interaction data in E-learning environments: the ELECTRE TRI method Download PDF

Şahin M. , Ulucan A., Yurdugül H.

EDUCATION AND INFORMATION TECHNOLOGIES, cilt.1, no.1, ss.1-18, 2020 (SSCI İndekslerine Giren Dergi)

  • Cilt numarası: 1 Konu: 1
  • Basım Tarihi: 2020
  • Doi Numarası: 10.1007/s10639-020-10358-2
  • Sayfa Sayıları: ss.1-18


E-learning environments can store huge amounts of data on the interaction of learners with the content, assessment and discussion. Yet, after the identification of meaningful patterns or learning behaviour in the data, it is necessary to use these patterns to improve learning environments. It is notable that designs to benefit from these patterns have been developed particularly with the use of educational data mining and learning analytics in the recent times. On the other hand, multi-criteria decision-making methods provide opportunities to researchers to discover and use the patterns in the data obtained from learning environments. This study seeks to discover the patterns in the interaction data gathered from e-learning environments. In this context, the research has two main objectives. Firstly, it aims to utilize the ELECTRE TRI method, which is one of the multi-criteria decision-making methods designed to classify the learners based on the interaction data in different units. Secondly, it aims to analyse the relationship between the classification based on the ELECTRE TRI method and the classification in the real life. To that end, two different interaction data sets obtained from learning management systems at different times are used in this study. The first data set consists of the data on 11 criteria and 78 students whereas the second data set consists of the data on 25 criteria and 65 students. Three different categories are identified in the first data set by the ELECTRE TRI method. Based on this finding, the classification in the ELECTRE TRI method is compared to the real-life classification, which shows a medium-level correlation. Two different categories are identified in the second data set. There is a medium-level correlation between these categories and the real-life classification as well. In conclusion, this study presents discussions on the use of multi-criteria decision-making methods to improve e-learning environments.