A Comparative model for predicting salary of Statistics and related field graduate from Thailand universities

Abstract

Thailand's graduate employment data have been collected continuously from the Ministry of Higher Education, Science, Research and Innovation. The data are collected from different universities in Thailand but it a huge amount of data, incomplete and high dimensional. An analysis of graduate employment, salary is one of the important factors that make graduates decide for work. This paper is to study a predictive model to predict the salary of Statistics and related fields graduate by comparing a model’s performance from statistical and machine learning models with RMSE and MAPE. The result got the most accuracy for predict graduate salary is a Random forest model but Random forest is complicated than linear regression to describe a model. Top 5 of importance variables from Random forest are Master level, Undefined job, Statistics major, Private job and Applied statistics major. Top 5 of importance variables from linear regression are Master level, Ph.D. level, Year 2017, Year 2016 and Time find work 1 - 2 months. The conclusions of the study are useful to help graduates decide a direction for their work.