Visualization of students' performance from digital learning media using Self-Organizing Map (SOM).

In recent years, with the increase of online learning platforms, obtaining learning behavior data from students, are becoming easy. However, analyzing the data to extract meaningful information remains challenging due to the data volume and complexity. Here, we apply Self-organizing Maps (SOM) to visualize the learning characteristics of many elementary students over many online assignments in mathematics class. As a case study, this research used the data log of MONSAKUN, a digital learning environment that focuses on exercising arithmetic using story-based questions by using a problem-posing approach with the integration of mathematical sentences. Our primary objective is to give intuitive understanding to the teachers regarding the students' performance that subsequently allows the teachers to generate meaningful advices. Here, SOM generates a two-dimensional map that preserved the topological order of high-dimensional learning characteristics data, in which students with similar learning characteristics are located close to each other, while students with significantly different characteristics are distanced from each other. We are interested especially in locating low-performing students, as they are the most important to be given advice by the teachers. By locating the low-performing students and other students in their vicinity on the map, the teacher may be able to use other students as references for improving the low-performing students' performances. The idea is to mimic the learning characteristics of the students designated as references. Due to the learning characteristics similarity, the low-performing students do not need to make drastic changes in their learning styles. It can be expected that by iterating this process over many assignments, the students' performances will gradually be improved. In this paper, we utilized a digital learning platform MONSAKUN, for learning arithmetic used in Japan. In this paper, we present our preliminary results in the form a learning-visualization of characteristics students of analytical and advisory tools and described our future goals for building a more flexible and general tool that can be deployed in Indonesia. [ABSTRACT FROM AUTHOR]