Research on a Trend Forecast Model of Time Series Data Based on Deep Learning

Inhalt:
The prediction of stock price trends is of great significance in the study of time series. Based on deep learning, this paper combines a convolutional neural network (CNN) for feature extraction and a long short-term memory network (LSTM) for prediction, and proposes a 3dCNN-LSTM stock price trend prediction model. The three-dimensional input tensors of the model are the technical indicators of the stock index, the correlation between the stock indexes and the historical information of the time series. Among them, this paper converts the traditional technical indicators for predicting stock prices into trend-deterministic data, and secondly uses distance correlation coefficients to sort the stock indices, and designs a CNN module that can extract three-dimensional input features, and sets up simulation experiments under different conditions to determine the optimal parameters of the model. Through comparison with other prediction models, the results of simulation experiments show that the 3dCNN-LSTM model has a better prediction effect and improves the F1 measurement value.