Prediction of Intention to Use Social Media in Online Blended Learning Using Two Step Hybrid Feature Selection and Improved SVM Stacked Model

Abstract

The development of Information and Communication Technology (ICT) along with the widespread availability of smartphones and internet connections at affordable prices, lead to the exceptional growth of social media (SM) use among all fields. The field of education that witnessed wholesome changes due to the pandemic in the form of online classes, online exams is also impacted by the rapid development of SM. This research predicts the students' intentions to use SM for learning in higher education and also tries to identify the underlying reasons for this intent. In this study, three classical machine learning (ML) classifiers- C5.0, random forest (RF) and support vector machine (SVM) have been used to predict the target variable (intent to use SM for education) using survey data collected from students studying in Indian higher educational institutions. The study also proposes a new ML classifier by combining RF and SVM. Findings of the study indicate that an individual's perceived risk, perceived ease of use, ease of communication, time and place flexibility, and gender are important predictors of the target variable. The newly proposed model's performance accuracy is 100 percent and outperforms the classical ML algorithms in many scenarios.