Gold price prediction in covid-19 era
Seng Hansun, Alethea Suryadibrata · Computational intelligence · 2021
As one of the most frequently traded commodities in the world, gold has been hugely impacted by the COVID-19 crisis. In this study, we try to apply a famous Deep Learning method for time series analysis, namely the Long Short-Term Memory (LSTM) networks, for future gold price prediction. However, rather than using a complex network architecture, we propose simple three layers LSTM networks that were trained on 4,219 training records and tested on 1,055 test records. We found that the Root Mean Square Error (RMSE) value for the prediction results is 39.94162, while the Mean Absolute Percentage Error(MAPE) value is 17.66144. Moreover, the R2 score of the prediction results could reach 97.242%, which is considered high and comparable with other more complex networks’ architectures available in the literature. © Muk Publications.