Gold Price Prediction Using an Ensemble of Random Forest and XGBoost

Abstract

This research paper presents an optimized ensemble approach for gold price prediction by combining the Random Forest and XGBoost algorithms. The proposed methodology incorporates a meta-model to enhance the ensemble's predictive performance. Using data from Google Finance (2013-2023), we train and test the individual models before leveraging the meta-model to fuse their predictions into a unified ensemble prediction. Evaluation using multiple performance metrics, such as MAE, MSE, RMSE, R2, MAPE and Max AE, demonstrates the superior predictive capabilities of our ensemble approach compared to the individual algorithms. This study contributes to the advancement of gold price prediction by leveraging ensemble learning techniques and showcasing their effectiveness in capturing the complex dynamics of gold market trends. The proposed approach holds significant potential for improving financial decision-making and risk management strategies in the domain of gold investments.