Classification of Student Learning Styles Using Artificial Neural Networks on Imbalanced Data

Abstract

The transformation of learning activities towards digital form since the COVID-19 pandemic can affect students' learning process. One of the factors that can affect this learning process is the learning style owned by each student. Learning patterns that are not in line with students' learning styles can influence their learning process. This study aims to identify students' learning styles based on data extracted from the Moodle Learning Management System (LMS). The research methods applied in this study include data collection by extracting data from Moodle LMS logs and classifying student learning styles using the Artificial Neural Network (ANN) algorithm. This study uses 310 log extraction data on the Moodle platform. The Isolation Forest algorithm was applied to this study to detect anomalies or outliers in the dataset. The data used in this study also has an unbalanced distribution of data per class. To prevent the performance degradation of the classifier model caused by the imbalance of data distribution, this study uses the SMOTE algorithm which can generate new synthetic data on minority class. This study combines three algorithms consisting of the Isolation Forest Algorithm for dataset management, the SMOTE Algorithm to solve the problem of data imbalance, and the ANN Algorithm to build a classification model. The model evaluation is carried out by considering the values of accuracy, precision, recall, and F1-Score to identify the reliability level of the produced model. Based on the research, this study produced a classifying model with an accuracy of 96%. The model produced in this study can be used to identify students' learning styles and as a reference for improving the quality of the teaching and learning process.

Keywords

• Machine Learning
• Imbalanced Data
• Learning Style
• Artificial Neural Network
• SMOTE