Analysis and Prediction of Gold Price using CNN and Bi-GRU based Neural Network Model

Abstract

In recent decades, gold has been one of the most sought-after commodities for long-term and short-term investments, as investors perceive gold as a hedge against unanticipated market occurrences. Gold can be purchased, stored, and is rarely utilized as a payment method. However, it is pretty simple to convert gold into cash in almost any currency. As gold is essential for maintaining value, investment, and national economic stability, it is undoubtedly vital to forecast the price of gold accurately. In this paper, we proposed a hybrid method for forecasting the price of gold based on the combination of 1D Convolutional Neural Networks (CNN) and Bidirectional Gated Recurrent Unit (Bi-GRU). Though CNN-GRU, CNN-LSTM, CNN-RNN based hybrid networks or, the individual CNN, Bi-GRU, and GRU provide satisfactory results, our proposed hybrid approach is more reliable since it outperforms other networks and achieves the MAE, MAPE, MedAE, RMSE, MSE, MSLE of 11.88, 0.67%, 8.41, 15.59, 242.90, $76.08\times 10^{-6}$ respectively and $\mathrm{R}^{2}$ Score of 93.56%.