Forecasting Stock Prices in the Resources Industry Sector Using Machine Learning Techniques

Abstract

This research aims to develop and evaluate the efficiency of a forecasting model for predicting stock prices in the resource industry sector. The study utilized secondary data related to factors influencing stock prices from January 1, 2016, to September 30, 2023. Data were obtained from three sources: the Bank of Thailand, the Stock Exchange of Thailand, and Investing.com. The analysis followed the Cross-Industry Standard Process for Data Mining and employed three machine learning techniques: Neural Network, K-Nearest Neighbor, and Support Vector Machine. Moreover, model performance was evaluated using multiple statistical metrics: R-squared (R²), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Root Relative Squared Error (RRSE). The results indicated that the Neural Network technique demonstrated the highest forecasting accuracy, with an MSE of 581,543.51, RMSE of 762.59, RRSE of 0.31, and R² of 91.20%. These findings suggest that the Neural Network model outperformed the other techniques and can serve as a reliable tool for forecasting stock prices in the resource industry sector, offering valuable support for investment decision-making and risk reduction.