Decision tree for investigating the factors affecting graduate salaries

In recent years, the rapid expansion of higher education has led to a surge in competitive pressure in the education market and has increased the excess in university-educated labor supply. Thus, the salary of graduates is a major societal issue. Over the past two decades, a number of studies have explored the factors that positively affect graduate salaries, and have constructed a complete theoretical scheme. Our studies explored these factors by analyzing data from the Taiwan Education Panel Survey and Taiwan Education Panel Survey and Beyond, which together comprise 1,303 variables. The following three types of model were investigated: the initial model, workplace model, and change model. Data mining for the three models was conducted using a regression tree, a type of decision tree method for analyzing big data. The major findings of this research are described as follows: (1) Three variables, educational level, firm size, and academic achievement, were selected in the initial model. (2) Four variables, firm size, job title, working hours, and academic achievement, were selected in the workplace model. (3) One variable, job title, was selected in the change model. In summary, the decision tree effectively determined that academic achievement positively affects graduate salaries.