Robust Logistic Regression for Graduate's Employability from Public Universities in Malaysia

Abstract

The graduate’s employability is studied to investigate its factors involving graduates from public universities in Malaysia for the year 2016. This topic has become a concerning topic in Malaysia as the number of graduates getting hired is lower than expected and does not commensurate with the total number of graduates produced every year. The investigation on graduate’s employability based on their profiles is done by utilizing a robust method due to the existence of outliers in the data set. The analysis shows that age, CGPA, discipline, educational stage, English grade, entry qualification, gender, state, sponsor and university types are factors that affect Malaysian public university graduates. Overall, robust logistic model fits the data with correctly classified 69.65% and produces area under the Receiver Operating Characteristics (ROC) curve with 0.6961 in separating the groups of employed and unemployed graduates.