Developing Self-Adaptive Data Analytics Models for Personalized Visual Insights in Education

Educational systems are undergoing a transformation to meet the diverse learning needs of students. Traditional one-size-fits-all teaching approaches often fail to address individual differences in learning styles, pace, and preferences. The integration of data analytics and visualization techniques offers opportunities to enhance personalization in education. By leveraging advanced machine learning and visualization models, tailored learning experiences can be designed, leading to improved engagement and performance. This research introduces the Visualization-Driven Sequential Temporal Analytics Model (VISTEM), a novel hybrid algorithm that combines t-Distributed Stochastic Neighbour Embedding (t-SNE) and Long Short-Term Memory (LSTM) networks. VISTEM is designed to analyze sequential student learning patterns and provide adaptive, personalized visual insights to enhance the educational experience. The proposed model integrates real-time data analysis with advanced visualization techniques, enabling educators and stakeholders to identify knowledge gaps, predict performance, and tailor learning strategies. To validate its effectiveness, VISTEM is compared against existing models such as Recurrent Neural Networks, Principal Component Analysis and K-Means Clustering using simulation metrics such as prediction accuracy, training time and engagement level prediction. Simulation results demonstrate that VISTEM outperforms traditional algorithms in terms of sequential pattern detection, real-time visualization responsiveness, and interpretability. The integration of interactive visual elements provides enhanced insights into student behaviour, facilitating data-driven decision-making in education. This study underscores the potential of hybrid algorithms in creating scalable, efficient, and adaptive data analytics systems tailored for modern educational environments.