Learning Behaviour Analysis of Open Education Learners Based on Clustering Algorithm

Open education platforms have become one of the most significant modes of learning in the digital era, offering learners the flexibility to access diverse educational resources. However, analyzing learner behavior in such open environments is challenging due to the massive scale of data, varying engagement patterns, and the sparsity of meaningful interaction features. Traditional clustering methods often struggle to capture the non-linear and high-dimensional structures in learner behavior data, leading to limited accuracy and interpretability. To address these challenges, this research proposes a hybrid framework combining Autoencoder and KMeans clustering for learning behavior analysis. The Autoencoder effectively reduces dimensionality and extracts deep latent features, while K-Means clusters the learners into meaningful groups based on their behavioral similarities. Publicly available datasets such as Open University Learning Analytics Dataset (OULAD) and MOOC are employed for implementation and evaluation. Preprocessing techniques including normalization, handling missing values, and feature engineering were applied to ensure data quality. Experimental results demonstrate that the proposed Autoencoder and KMeans model significantly outperforms conventional clustering techniques in terms of silhouette score, accuracy of 98.4 %. The research highlights the potential of deep learning and clustering methods to provide actionable insights for personalized learning, student retention strategies, and improving the overall effectiveness of open education systems.