Stock index prediction system based on CEEMD-GRU model

Abstract

The invention belongs to the technical field of pattern recognition, and provides a stock index prediction system based on a CEEMD-GRU model, the stock index prediction system is provided with a model storage part, a preprocessing part and a result prediction part, the model storage part stores the CEEMD-GRU model, the preprocessing part can preprocess original stock data to improve the data quality, the smoothness of the model training process and the final fitting effect are guaranteed, and a stock index prediction result can be obtained by a result prediction part. In the training process of the CEEMD-GRU model, the processed data is subjected to modal decomposition through a CEEMD algorithm, then components are given to each independent GRU model for training and iteration, the GRU model can be used for independently learning different time scale features, the learning effect on different features is guaranteed, and the learning efficiency is improved. Therefore, the purpose of overall grasping and learning of original data features is achieved.

Description

technical field [0001] The invention belongs to the technical field of pattern recognition, and in particular relates to a stock index prediction system based on the CEEMD-GRU model. Background technique [0002] Stock forecasting has always been a hot research field, and there are many research methods for it, which can be divided into two categories: fundamental analysis and technical analysis. The fundamental analysis method focuses on macro analysis, and realizes the value evaluation of stocks based on factors such as the overall economic situation, policies and regulations. The fundamental analysis method has a high theoretical threshold and a large research cost for ordinary investors, and does not have the advantage of universal application. [0003] The rules of technical analysis are based on the securities market itself, and use many technical indicators to study the characteristics and laws reflected in the past and present of the stock market, so as to achieve t...

AI Technical Summary

Problems solved by technology

However, if the original data is directly input without any processing, or the modeling method only relies on a single model, the fitting effect on the real data is relatively limited, resulting in low accuracy of stock forecast results.

Images