USE OF DATA MINING BASED ON ACADEMIC SYSTEMS FOR STUDENT PERFORMANCE MONITORING AND ATTRITION
Attrition, Retention, Moodle,AVA (Ambiente Virtual de Aprendizagem): VLE (Virtual Learning Environment), Knowledge discovery
This work is the qualification for the Professional Master's in Informatics in Education at the Federal Institute of Science and Technology of Rio Grande do Sul - Porto Alegre campus. Its focus is the analysis of students' academic life to assess the risk of attrition, retention, dropout, and academic performance, using historical and current data from academic systems and the Moodle Virtual Learning Environment (AVA). It provides knowledge mining through prediction and description algorithms usable in educational management initiatives aimed at improving efficiency. It also enables inference between the methodology used in virtual environments and performance on these platforms.
To achieve the desired results, a methodology derived from CRISP-DM divided into 10 project flow management phases will be applied. The MariaDB DBMS will be used for the research development phase, and either Spark or Hadoop MapReduce will be chosen for the subsequent presentation of the results obtained. Three different tools with the same purpose will be used for algorithm execution, which are WEKA, Orange, and Knime.
The obtained results should be divided into two main parts. The first part includes the minimal data for identifying the student at the institution, course, and classes, while the second part provides the mined knowledge that translates the assessed risks regarding attrition, retention, performance, and dropout. The availability of the results for reading should be implemented with security authentication through web services, facilitating the collection of acquired knowledge and allowing the construction of informational dashboards and other applications that can utilize this knowledge.