A Data Mining Approach to Construct Classification Model for Predicting Tourism Graduates Employability

Abstract

Higher educational institutions’ goal is to ensure that graduates are imbibed with the required employability skills. Hence, by applying the concept of Knowledge Discover in Database (KDD), this study aims to build a graduates’ employment prediction model using classification task in Bayes and Tree Methods. The data utilized for this purpose are collected from the tracer survey conducted to Oman Tourism College alumni. Based on the graduates’ profiles, the generated model predicts whether a graduate is employed full-time, part-time, self-employed or unemployed. Using several classification techniques provided by Waikato Environment for Knowledge Analysis (WEKA), the findings revealed that RandomTree algorithm and REPTree algorithms, under decision tree methods yielded accuracy rates of 96.3636% and 88.1818% respectively. BayesNet algorithm, a variant of Bayes algorithm yielded an accuracy of 84.5455%, ranked third. Information gain and ranker method are also used for attribute ranking which showed occupation as the most influential factor for employability followed by job sector. Other attributes used in classifying the employment status of the graduates include occupation, job sector, specialization, degree, age, personality development skills, cultural competency, leadership, interpersonal skills, creativity, and problem-solving skills. This experiment concludes that a tree-based classifier is the most suitable algorithm for predicting tourism graduates’ employability in the tourism and hospitality sector of the Sultanate of Oman.