Development of a CNN-LSTM Approach with Images as Time-Series Data Representation for Predicting Gold Prices

Abstract

Predicting gold prices is not easy due to its non-linear, unpredictable, volatile, and uncontrollable price movements. In this research, a combination of convolutional neural network (CNN) and long short-term memory (LSTM) is used to predict gold prices. By combining these two methods, the prediction model can leverage the strengths of CNN and LSTM to improve accuracy and learning performance. In addition, this CNN-LSTM model is enriched with input in the form of images that represent the timeseries data, where the gramian angular field (GAF) technique is used in timeseries data to images transformation. Experimental results showed that the proposed approach performs significantly better compared to the benchmark model.