CONSTRUCTION METHOD FOR FOREIGN EXCHANGE TIME SERIES PREDICTION

摘要

Disclosed is a construction method for a foreign exchange time series prediction, relating to the field of foreign exchange time series data. In the method, foreign exchange time series data is analyzed and predicted by using a deep learning algorithm C-LSTM, which combines a convolutional neural network with a long short-term memory network. A construction method for a network structure comprising five functional modules, comprising an input layer, a hidden layer, an output layer, network training and network prediction, is proposed. The construction method comprises: selecting an activation function of the C-LSTM, which combines the convolutional neural network with the long short-term memory network; defining a loss function of the C-LSTM, which combines the convolutional neural network with the long short-term memory network; and selecting transaction-type indicators and fundamental data to be input features of the C-LSTM, which combines the convolutional neural network with the long short-term memory network. In combination with the advantages of convolutional neural network and long short-term memory network algorithms, the construction method for a foreign exchange time series prediction is proposed. On the basis of the construction method, temporal and spatial features of foreign exchange time series data can thus be better analyzed and mined.