Abstract:
"
The gold price variation is a major concern for investors, gold buyers and sellers, gold mining
establishments and government since price of gold has significant impact on financial and
economic activities take place throughout the world. The development of reliable forecasting
system has the capability to offer the insights in gold price fluctuations and capture the price
change and ultimately provide opportunity to gain profits and limit losses. Nevertheless, its
challenging effort due to the multi influence factors and non-linearity nature in gold market.
The author has made use of deep learning approach forecasting model for accurate
forecast of gold price. In this research study, Long Short-Term Memory (LSTM) Network with
hyperparameters tuned used for the daily gold price forecasting. Several highly correlated
predictor variables of gold price were used as multivariate inputs to build LSTM for forecasting
gold price with the use of the dataset of daily prices over the period of January 2014 to March
2021 from Yahoo Finance.
Author conducted a sequence of trials and evaluated the proposed model opposed to
the state of art machine learning and deep learning models. The experimental results exhibit
that the proposed LSTM model has a magnificent performance in forecast with lowest Mean
Absolute Error (MAE) and Root Mean Square Error (RMSE) and highest Coefficient of
Determination (R2
) score. The proposed forecasting model distinguished as promising
technique for gold price forecasting with experimental results."
