Educational data mining in a university using LMS Moodle

Educational data mining is a technology for processing big data collected through educational portals and used for assessing the quality of educational content and predicting educational outcomes. Various data analysis methods are applied in the implementation of this technology, including anomaly detection, association rule mining, classification, cluster analysis, regression analysis, and factor analysis. Each of these methods uses special tools, such as spreadsheets, high­level programming languages (Python or R), low­code platforms (Loginom, Orange), or specialized analytics modules integrated into educational portals. The article reveals the features of the “Analytics” module embedded in LMS Moodle for tracking and optimizing the educational process. The authors provide recommendations for configuring and using this module in the educational process. The module “Analytics” allows to create forecasts of the success rate of mastering a separate training course without programming skills , does not require data preprocessing for analysis, and allows to quickly take measures to improve the educational process based on the results of the forecast. However, the module has a number of disadvantages (limited attributes and forecasting models, high requirements for computing power to run model training and build forecasts), so often administrators of educational portals disable this module. In this regard, an experiment was conducted to train a classification model for predicting students’ examination scores based on students’ activities on the educational portal of Nosov Magnitogorsk State Technical University. In the course of the experiment, data on students’ activity on the university’s educational portal for one academic year were taken, and machine learning models were built to predict examination scores for an individual student and subject. The results of the experiment showed F1 74 % in predicting the score that the students would receive on the intermediate assessment. This confirms the effectiveness of using educational data mining based on user data from educational platforms.