Predicting Engineering Students' Employability Using Data Mining Classification Techniques

Higher education institutions in Oman emphasize the employability of their students as one of their top priorities. Hence, the primary objective of this study is to construct a model to predict engineering students' employability using data mining classification techniques. It also investigates relevant factors to improve the prediction model. The study applied a technique known as Knowledge Discovery in Databases (KDD), which utilized 250 engineering students' records stored in the University of Nizwa database from the Spring Semester of the Academic Year 2020–2021. WEKA-Waikato Environment for Knowledge Analysis- was used to create the research classifiers. The InfoGainAttributeEval and the Ranker search methods were used to identify the most significant attributes of student employability. After the testing phase, 10-fold cross-validation was carried out. The prediction models J48 and RepTree outperformed Random Tree in accuracy metrics and tree size. J48 showed the highest Kappa value, lower error rates, and compactness, making it the top choice. Attributes like internship mark, credit hours, department, and specialization heavily influenced model performance, with J48 exhibiting superior accuracy and interpretability over RepTree. Using the most accurate data mining classification algorithm, a prediction model was developed as a subsequent stage of the data mining technique. The study suggested recommendations to improve the student's employability by emphasizing skills and competencies related to employment.