A comparative of predictive model of employability

Abstract

In 2017, the global unemployment rate is projected around 5.6% while for 2018 the unemployment rate is 5.5% which is little bit decrease. However, the youth (aged 15 to 24) unemployment rate in Malaysia is over three times higher at around 10.8% in 2017. In addition, Malaysia achieved the second highest rate after Indonesia (15.6%) compare to other countries in Asian including China (10.8%), India (10.5%), Singapore (4.6%), Vietnam (7%), Thailand (5.9%) and Philippines (7.9%). This study aim to present a set of data mining algorithms to find the most important factor of employability among the fresh graduate students. The comparison for six data mining algorithms which are 1) Logistic Regression, 2) Decision Tree, 3) Naive Bayes, 4) K-Nearest Neighbor, 5) Support Vector Machine and 6) Neural Network by using split validation method which is 70-30 as a ratio. Based on the result, Neural Network is the best classifier other than another five algorithms. The Neural Network Model showed 6 majors effect on employability are 1) willing to face challenges of the outside world and work, 2) can communicate effectively, 3) field of technical, 4) convocation on October and 6) Sex (Male). The predictive model of employability will benefit the management of the higher education, Ministry of Education and fresh graduate itself to predict the employability status either employed and unemployed by graduate data.