A Robust Prediction for Gold Price using Heterogeneous Ensemble Learning

Nowadays, gold price prediction in financial markets holds significant implications for investors, policymakers, and industries worldwide. In previous studies, gold prices were often predicted by single machine learning models, which did not combine the strengths of different algorithms. In this study, we propose a robust prediction model for gold prices employing heterogeneous ensemble learning (HEL). Our approach combines algorithms such as linear regression, random forest, gradient boosting, support vector machines, and Gaussian process regression in the first-level learner, with XGBoost in the second-level learner. The study provides a comprehensive evaluation of the proposed HEL model by comparing it to single machine learning algorithms and a deep learning model, Long Short-Term Memory (LSTM). Metrics such as mean absolute error (MAE), and mean absolute percentage error (MAPE) are used to evaluate the models. Our results reveal that the HEL model outperforms individual machine-learning algorithms. Furthermore, it demonstrates greater effectiveness when compared to LSTM, a powerful deep-learning model.