Model Prediction for the Number of Foreign Workers using Data Mining Techniques and Statistical Analysis

Abstract

This research aims to study the factors that affect the entry of foreign workers in Thailand and to create an appropriate statistical model for forecasting the number of foreign workers in Thailand. With multiple linear regression, decision tree, and support vector regression methods using information from the foreign workers in Prof. 2011 – 2021 for 132 months, there are a total of 9 related factors, including the minimum wage, the unemployment rate, the gross domestic product value, the effective exchange rate, the inflation rate, the export value, the import value, the number of Thai workers authorized to work abroad and the number of foreign workers in the past unit of time. The research process is carried out in accordance with the 5 steps standard procedures for data mining (CRISP-DM). To analyze related factors, the statistical package is used, and the effectiveness of the model will be tested using the WEKA program. There are 3 criteria for measuring the performance of the prediction: Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and Coefficient of Multiple Determination (R2). The results showed that factors affecting the entry of foreign workers in Thailand are the minimum wages, the inflation rate, the unemployment rates, the gross domestic product value and the number of foreign workers in the past unit of time. The performance measurement of the model found that the support vector regression methods had the lowest MAE and RMSE, 15.7566 and 7.7770, respectively, and had R2 of 0.8847, which is the 2nd largest after multiple linear regression. Therefore, support vector regression technique is the most effective and most suitable for forecasting the number of foreign workers in Thailand.