Multivariate time series prediction method and system, computer product and storage medium

Abstract

The invention discloses a multivariate time series prediction method and system, a computer product and a storage medium, and the method comprises the steps: respectively extracting spatial and temporal feature vectors of long and short term historical data matrixes through employing two feature extraction codes, inputting a historical time series matrix into a spatial feature extraction encoder, generating a weighted attention spatial feature vector, and carrying out the prediction of the spatial feature vector; inputting the weighted spatial feature vector into a gating cycle unit to generate a spatial-temporal feature vector; inputting spatio-temporal feature vectors extracted from the long-term historical data matrix into an interactive attention module to generate weighted feature vectors; inputting the short-term historical data matrix into an autoregression layer, and generating a linear prediction result of the short-term historical time sequence data; and combining and inputting the weighted feature vector and the coding feature vector into a full connection layer to generate a neural network prediction result, and adding the neural network prediction result and an autoregression layer linear prediction result to obtain a final prediction result. According to the invention, accurate prediction of multivariate time series data is realized.

Description

technical field [0001] The invention relates to the field of multivariate time series data prediction, in particular to a multivariate time series prediction method and system based on feature extraction coding and interactive attention modules. Background technique [0002] With the development of big data technology and the rapid growth of data, using time series data to predict its future state has a wide range of application scenarios, such as traffic flow prediction on traffic lines, stock price prediction on the stock market, Air Quality Index Forecast. Accurately predicting new trends or potential events is often what users are really interested in, and provides strong support for future decision-making and planning, and is helpful for the implementation of advanced applications. However, problems such as complex periodic patterns and dependencies in time series data cannot be well modeled. The industry has conducted in-depth research on the above issues. [0003] ...

AI Technical Summary

Problems solved by technology

[0023] The technical problem to be solved by the present invention is to provide a multivariate time series forecasting method, system, computer product, and storage medium for the deficiencies of the prior art, so as to overcome the inability of the prior art to learn complex periodic patterns, time and space dependencies , the problem of low prediction accuracy, and realize accurate prediction of multivariate time series data with dynamic changes in space-time relationship

Images