Time series data prediction method, device and equipment

Abstract

The invention discloses a time series data prediction method, device and equipment. The method includes: acquiring historical time series data, and subjecting the historical time series data to data cleaning and data slicing to obtain a corresponding time series data sequence; subjecting the time series data sequence to stabilization, and adopting an immunogenetic feature reconstruction algorithmfor feature reconstruction of the stabilized time series data sequence to obtain a corresponding feature sequence; acquiring a deep learning model based on training of the feature sequence, and adopting the deep cleaning model for time series data prediction. Different from the prior art in which data set feature acquisition is realized according to a sampling method, the time series data prediction method guarantees effectiveness of the acquired time series data features by steps of data preprocessing, stabilization, feature reconstruction and the like, and accordingly the deep learning modelis enabled to learn the time series features of the time series data, and prediction accuracy of the deep learning model is guaranteed.

Description

technical field [0001] The present invention relates to the technical field of deep learning, and more specifically, to a time series data prediction method, device and equipment. Background technique [0002] "Exchange rate" is referred to as ExRate, also known as "foreign exchange rate", "foreign exchange market" or "exchange rate", etc. It is the ratio of one currency to another currency, reflecting the change in the relative price of the currency; with the global floating exchange rate With the legalization of the system and the strengthening of the trend of world economic integration, foreign exchange, as an important capital element, has become an important component of many capital products. Therefore, its prediction has attracted the attention of investors from all walks of life and the securities market. [0003] In recent years, with the development of heuristic algorithms, various machine learning algorithms have been applied to exchange rate forecasting. Specific...

AI Technical Summary

Problems solved by technology

[0003] In recent years, with the development of heuristic algorithms, various machine learning algorithms have been applied to exchange rate forecasting. Specifically, the generalized autoregressive neural network (GRNN) model has been proposed in the prior art to simulate nonlinear changes in exchange rates, but This kind of static network model needs to learn the characteristics of the input data set through random sampling, which makes the model unable to learn the time series characteristics of exchange rate changes

[0004] To sum up, there is a problem in the technical solutions used to realize exchange rate forecasting in the prior art that the model cannot learn the time series characteristics of exchange rate changes

Images