Performance Analysis of Data Classification Algorithms in Online Social Networks

Predicting student performance in online social networks (OSNs) is a complex task that involves analyzing various factors, including students' online behaviour, interactions, and engagement patterns. By leveraging Machine Learning (ML) and Deep Learning (DL) techniques, researchers can develop predictive models that identify potential factors influencing student performance. This paper presents a comprehensive comparative study on the classification of real-time OSN user datasets using various machine learning and deep learning approaches. A three- methodology is proposed, introducing novel methods in each method to improve classification accuracy and efficiency. The first method combines Improved Mutual information based Filter Pearson’s Correlation (IMIFPC) feature selection with Enhanced RandomBayesian classification method (IMIFPC-ERB), Second method integrates Hybrid Convolutional Neural Networks and Long Short-Term Memory networks (HCNN-LSTM), and the third method leverages Recurrent Neural Networks and Bidirectional Long Short-Term Memory Networks (HRNN-BILSTM). Experimental results demonstrate the superiority of the HRNN-BILSTM algorithm, achieving improved accuracy, precision, recall, and F1-score compared to existing models. This study highlights the effectiveness of hybrid approaches in improving classification accuracy and efficiency for real-time OSN user classification.