Predicting Student's Soft Skills Based on Socio-Economical Factors: An Educational Data Mining Approach

Abstract

Recent changes in the labor market and higher education sector have made graduates' employability a priority for researchers, governments, and employers in developed and emerging nations. There is, however, still a dearth of study about whether graduate students acquire the employability skills that businesses want of them because of their higher education. To determine a student's future employment and career path, it is critical to evaluate their soft skills. An emerging area called educational data mining (EDM) aims to gather enormous volumes of academic data produced and maintained by educational institutions and to derive explicit and specific information from it. This paper aims to predict students' soft skills such as professional, analytical, linguistic, communication, and ethical skills, based on their socio-economic, academic, and institutional data by leveraging data mining methods and machine learning techniques. All five soft skills were predicted using prediction models created using linear regression, probabilistic neural networks, and simple regression tree techniques. This study used a dataset from an open source that Universidad Technologica de Bolivar published. It covers academic, social, and economic data for 12,411 students. The experimental results demonstrated that the linear regression algorithm performed better than the others in predicting all five soft skills compared to machine learning methods. This finding can assist higher education institutions in making informed decisions, providing tailored support, enhancing student success and employability, and continuously modifying their programs to meet the needs of students.