Customized support vector machine for predicting the employability of students pursuing engineering

Higher Education Institutions are always concerned about their students’ employability, so it is ideal to have a mechanism to forecast students’ employability at an early stage so as to take appropriate action to improve the same. There could be many factors that can have a direct or indirect affect on employability. In this research, we have analyzed the students data who have graduated  between the years of 2018 and 2022  from an engineering college and we considered both academic and socio-demographic factors. To identify the best attributes that impact employability, we have used our proposed feature selection approach, which is influenced by the Teaching Learning Based Optimization (TLBO) algorithm. The study employs several classifiers, including Random Forest, Decision Tree, K-Nearest Neighbor, Support Vector Machine, Gradient Boosting Adaptive Boosting, and Extreme Gradient Boosting and found that Support Vector Machine(SVM) provides the greatest accuracy of 74.37%. Then to improve the accuracy of SVM we have proposed Customized SVM where we have customized the hyperparameters like the kernel function’s radial basis function and regularization parameter, C value. We found that with this novel approach, accuracy improved by 13.43%. Thus this proposed novel approach helps in finding an optimal set of features and the best classifier for predicting students’ employability enrolled in a Technical Higher Education Institute.