An Accurate Salary Estimation Scheme by Using BigData Technique

A suitable salary can benefit both the company and employees for higher company benefits and employee living quality. Most of the existing models focus on how to calculate employees' salaries by using employee personal information or how to eliminate bias in the machine learning process. To care more about employees' satisfaction and living costs, our salary prediction model is developed for salary estimation by analyzing the living cost and employees' background for both employers' and employees' satisfaction. However, An experiment is also conducted to compare three different model performances including Random Forest (RF), Neural Network (NN), and Support Vector Regression (SVR). The parameters of these models have been adjusted to get better accuracy. RF gives the best performance of Mean Absolute Error (MAE) at 829.688, while the NN with 4 layers got a higher MAE at 1087.115. The SVR method gets relatively poor performance for the 1471 records training datasets using the 3-fold cross-validation method. Finally, we select the RF model as our salary-estimating scheme and the linear regression model as our living-cost estimating scheme to build a salary-estimating system with a Web-based graphical user interface.