Abstract:

Developing teaching methods, designing assessments, analyzing educational data, and using the resulting insights to improve learning and teaching are the subject of much current research. In education, Learning Management Systems are used in various forms of learning: traditional learning supplemented by information and communication tools, blended learning, and distance learning. A Learning Management System combines a set of functions that allow the teacher to perform activities in an online environment (providing learning materials, communicating with students, organizing e-activities, assessment). They provide data on learner activities, such as click-based data describing whether, when, and how often learners access resources that provide different views of content, as well as data reflecting learner activity in the course. This data is available in data-driven reports embedded in the e-learning system, but these reports are often primarily descriptive, telling learners what happened but not why it happened, and they do not predict outcomes or give learners advice on how to improve their outcomes.

Data mining is used in education to identify learning problems, investigate and predict learner performance, and evaluate the integration of technology into the learning process. This paper presents a method for identifying predictive factors for student failure in STEM oriented course using Decision Trees and Logistics Regression. The independent variables used are the number of instructional and self-study materials downloaded, the points earned on weekly quizzes, the number of video lessons viewed, and the total student activity in the e-course, and the dependent variable is the number of points obtained in the midterm exam. The obtained results are used to further improve the teaching process in the future.

Keywords:

Data mining, Decision Trees, Logistic Regression, identification, student failure.