The analysis of learning style and strategies used in e-learning environment via web usage mining

Regulation of e-learning environments considering students? individual differences is an important subject of educational technologies. Preparation of appropriate learning environments for individuals by identifying learning styles or learning strategies with the help of scales used at the beginning of student learning process is not an adequate approach. But during the learning process, the repetition of scales can lead to negative approaches among students. For these reasons, it is necessary to detect students' learning styles and strategies from behaviours in e-learning environment during the process with the aid of artificial intelligence systems. Thus, the activities offered to individuals in the teaching process will be regulated according to individual differences. In the design of these kinds of e-learning environments, determination of students' learning styles and strategies with artificial intelligence models via the web usage behaviour constitute an important dimension.In this study, it is aimed to analyze learning style and strategies used in e-learning environment via web usage mining. The study is designed according to a single group pretest-posttest pattern among experimental research models. When the results of learning styles and learning strategies scales applied to students' within the scope of the research are assessed, It is understood that some changes have been occurred in students' learning styles and strategies during the period they use e-learning environments. We observe some differences between the distributions of the pretest and the posttest of learning styles used by the students'. In addition, there is a significant (p<0,05) change between the dimension of abstract conceptualization and concrete experience which constitute learning styles, in terms of mean scores there is not a significant change between active experience and reflective observation but we observe a change between them. According to results obtained from motivation and learning for strategies scales, pretest-posttest differences in total scale scores aren?t statistically significant, but some significant differences between subscales of scale have been identified. In order to determine the relationship between learning styles and learning strategies, the correlation analysis for pretest and posttest scores has been separately performed between the dimensions of learning styles and motivation and learning for strategies. As a result of the analysis, a significant relationship has been observed between active experience and reflective observation of learning style dimension and sub-dimensions of motivation and learning for strategies, but determination coefficient (r2=%8) is found to be low.Artificial Neural Networks (ANN) and Decision Tree models are applied to determine the learning styles and strategies from students' web usage behaviours with the help of SAS Enterprise Miner 7.1 program. Visiting times for the activities and numbers of 10 learning activities presented for students in e-learning environment consist the web mining models? inputs. When the results obtained from Decision Tree and ANN models are evaluated, Decision Tree model is understood to be more successful than the ANN model. ANN model is evaluated in terms of their algorithms, Levenberg-Marquardt algorithm 91.7% is found to be successful to predict the students' learning styles.Data mining models have been tested for the learning strategies, but as both total scale scores and sub-dimensions of motivation and learning for strategies scale are used as the ANN model?s output, artificial intelligence training is found to be unsuccessful.When the results of this research have been evaluated, students' learning styles have been successfully predicted 91.7% from students? web usage behaviour with the help of artificial intelligence model obtained from this research. Artificial intelligence models can be used in the preparation of web-based intelligent learning environments and semantic web applications infrastructure. This artificial intelligence model can be used for the infrastructure of systems that can perceive the changes in learning styles of students in the learning process and offer learning environment appropriate for individual differences.