Hybrid RNN-BiLSTM Approach for Student Success Estimation in Online Social Networks

Abstract

Predicting student academic performance is a crucial task in educational data mining, enabling early identification of at-risk students and informed instructional interventions. This paper proposes a hybrid RNN-bi-LSTM method, combining the strengths of recurrent neural networks (RNNs) and bidirectional long short-term memory (bi-LSTM) networks, to model student learning behaviors and predict student performance. The hybrid approach leverages sequential dependencies and contextual information to improve prediction accuracy. Experiments on real-time high dimensional OSN user dataset demonstrate the effectiveness of the hybrid RNN-bi-LSTM method, outperforming baseline models in predicting student performance of Excel and Vivekanandha classes using neural network methods was tested. This study contributes to the development of accurate and reliable student performance prediction models, supporting data-driven decision-making in education and enhancing student success. The proposed hybrid classification performance strategy yielded an accuracy of 99.87 percent and an F1-score of 99.25 percent, according to the experimental findings obtained from the OSN User dataset.