Spatio-temporal data prediction method capable of directly capturing spatio-temporal correlation

Abstract

The invention discloses a spatio-temporal data prediction method capable of directly capturing spatio-temporal correlation. The spatio-temporal data prediction method comprises the following steps: preparing data required by the method; constructing a spatio-temporal data prediction model capable of directly capturing spatio-temporal correlation; training the constructed model to obtain model parameters; and predicting the spatio-temporal data by using the trained model. According to the method, the defect that an existing spatio-temporal data prediction method cannot capture tight spatio-temporal correlation in spatio-temporal data is overcome, spatio-temporal features in the spatio-temporal data can be effectively mined, so that the accuracy of a spatio-temporal data prediction task is improved, and the method has very important application value in the fields of traffic management, disaster prediction and the like.

Description

technical field [0001] The invention relates to the field of time series forecasting, in particular to a spatio-temporal data forecasting method which can directly capture the spatio-temporal correlation. Background technique [0002] Spatio-temporal data is the intersection of temporal data (time series) and spatial data. It refers to the data collected by entities or events with spatial coordinates at different times. It can be understood as a collection of time series generated on different spatial coordinates. Real life is full of various spatiotemporal data, such as traffic monitoring data, regional disaster data, etc. [0003] Time series forecasting refers to analyzing the development process, direction and trend of the obtained data in chronological order, so as to predict the possible development of several periods in the future based on past trends. Spatio-temporal data forecasting is a special case of time series forecasting, which is mainly reflected in the fact...

AI Technical Summary

Problems solved by technology

[0004] Most of the existing spatio-temporal data prediction methods ignore the spatial structure information in the spatio-temporal data, and regard it as a simple time series for prediction; or split the spatio-temporal data into two parts, time and space, and extract their temporal and spatial features respectively , ignoring the tight spatiotemporal correlation in spatiotemporal data

Images