Educational data mining to detect key resources

Abstract

This article describes an educational innovation project whose aim is designing and developing a new Moodle block to obtain decision tree based predictive models from usage data store in Moodle. Due to the features of these decision tree based models, it is possible to establish a relationship between student’s work and the final mark, and to obtain a description of the key resources to pass/fail a certain subject. With this information, the instructor could detect students with a high probability of fail and then try to correct the situation. This block could be applied to any subject in the Moodle platform. After developing the block, it has been tested with some of the subjects given by teachers involved in the project.