Gold price prediction using long short-term memory approach / Ahmad Ashhar Selamat and Anis Amilah Shari

Gold price prediction is crucial for investors and traders to make informed decisions and mitigate financial risks. This research project focuses on developing a robust gold price prediction system using the Long Short-Term Memory (LSTM) approach. The need for accurate and reliable gold price predictions is emphasized, considering the dynamic nature of the gold market and its significant impact on the global economy and financial markets. The proposed LSTM model aims to capture both short-term and long-term dependencies in time- series data, providing valuable insights for stakeholders in the gold market. The study addresses the complexity of gold price prediction and the limitations of traditional forecasting methods. By leveraging LSTM, the model is designed to effectively capture historical price data and relevant variables, offering a promising solution to the challenges faced in predicting gold prices. The methodology involves training and evaluating the LSTM model using historical data from reputable sources, ensuring the optimization of standard hyperparameters to achieve the best possible results. The results of the LSTM model demonstrate its superior performance, surpassing other machine learning models in terms of accuracy and reliability. The model exhibits a high level of accuracy of 96.9% With an impressive MAPE value of 0.031, showcasing its potential for practical application in the gold market. The findings highlight the effectiveness of LSTM in gold price prediction and underscore the significance of leveraging advanced machine learning techniques for commodity price forecasting. In future work, the system will undergo further enhancements, potentially incorporating adversarial learning to evaluate the robustness of the model. This ongoing development aims to continually improve the performance and reliability of the LSTM-based gold price prediction system, ensuring its relevance and effectiveness in real-world scenarios.