Development of Machine Learning Models Using Study Behavior Predictors of Students' Academic Performance Through Moodle Logs 23

Abstract

European Union project under the ESPRIT funding initiative in 1997 [7]. It is also commonly described this model as both a process and a methodology. As a methodology, it includes descriptions of the typical phases of a project, the tasks involved with each phase, and an explanation of the relationships between these tasks. As a process model, it provides an overview of the data mining life cycle. The researcher selected this process model over the other existing models because its framework has been anchored on business understanding (data mining goals) that other models failed to emphasize. The process or methodology of CRISP-DM contains six major steps namely: (1) business understanding, (2) data understanding, (3) data preparation, (4) modeling, (5) evaluation, (6) deployment [8]. Figure 1. Predictive Analytics Process Model Figure 1 indicates that the process model starts by gathering data of interest from the data lake (which is the Moodle database in this study). In an iterative process, features from the Moodle logs are selected using feature selection techniques of Weka such as CFS Subset Eval and then trains the models based on the predictors that are correlated to students' academic performance. Data mining techniques, statistical methods, and machine learning algorithms are used to determine the best features and the best models. Once an appropriate feature set and model are generated, thresholds are set to determine prediction results. The model/s with best predictive accuracy will be used and are then put into production to begin producing predictions about future data. Predicted results generated by the model/s are reflected back to Moodle through an integration table and Moodle block developed by the researcher.