Prediction of Student Employability through Internship based on Big Data Analysis

Abstract

The internship course is among the most important because it gives students a hands-on opportunity to apply their knowledge and get ready to launch a professional career. Internships, yet, don't ensure employability, particularly in cases where a graduate's performance is subpar and internship requirements are not fulfilled. Researchers in the field of higher education face a significant challenge in predicting employability due to the multitude of factors that contribute to this issue. This research presented the methodology for more accurately classifying student data in order to overcome this drawback. Online surveys were used to collect data from Princess Nourahbint Abdulrahman University's College of Computer and Information Sciences information systems (IS) graduates (PNU). The Switching Hierarchical Gaussian Filter (SHGF) is used to preprocess the data at the pre-processing stage. The outcome from the pre-processing is transferred to the feature selection method, which uses Siberian Tiger Optimization (STO) to select the student features. The employed, continuing studies, unemployed, and training are successfully classified using the Multi Fidelity Deep Neural Network. The proposed MFDNN -STO applied to the MATLAB/Simulink platform. To calculate the proposed approach, performance metrics including recall, ROC, computation time, accuracy, precision, sensitivity, and F-score were examined. Higher accuracy of 16.65%, 18.85%, and 17.89%, as well as higher sensitivity of 16.34%, 12.23%, and 18.54%, are achieved by the suggested MFDNN-STO method. The computation time was reduced by 14.89%, 16.89%, and 18.23% as well as 82.37%, 94.47%, and 87.76% in comparison to the existing method.