Using M Tree Data Structure as Unsupervised Classification Method

Increasing the effectiveness of educational processes is one of the greatest challenges for information society. The paper presents the usage of M Tree structure for classification of the learners based on their final marks obtained in their respective courses. The classical building algorithm of M-Trees with an original accustomed clustering procedure was implemented. The data that are managed within M Tree structure are represented by instances. The main goal of the structure is to provide information to students and course managers regarding the knowledge level reached by students. The proposed clustering procedure that is used for splitting full M Tree nodes is designed to properly classify learners. A baseline classification scheme based on k-means clustering and a custom M Tree clustering are presented. For comparison, there are considered classical characterization formulas.