Prediction model construction method and device and real-time prediction method and device

Abstract

The present disclosure relates to a prediction model building method and device and a real-time prediction method and device. The predictive model construction method includes: identifying a variety of different association patterns in the data to be trained, wherein a variety of different association patterns describe a variety of different association relationships between the influencing factors in the data to be trained and the target data; using Multiple sets of data corresponding to a variety of different association modes are trained separately to obtain multiple sub-prediction models suitable for a variety of different association modes; and building association modes between various different association modes according to the data to be trained A transfer model, wherein the association mode transfer model is used to determine the matching degree between multiple different sub-prediction models and the data to be predicted during the prediction process. According to the embodiments of the present disclosure, multiple prediction models corresponding to various data characteristics can be obtained, so the weight of each prediction model can be dynamically adjusted based on the characteristics of the data itself during prediction, thereby improving prediction accuracy.

Description

technical field [0001] The present disclosure relates to the technical field of prediction, and more particularly, to a method and device for building a prediction model and a method and device for performing real-time prediction. Background technique [0002] Prediction technology is a hot research issue at present. Forecasting technology can be widely used in various fields, such as air quality forecasting, climate forecasting, traffic flow forecasting, etc. According to the characteristics of the data and the different forecasting problems, various forecasting models have been proposed, such as parameter-based time series forecasting, neural network, regression analysis, etc. [0003] Traditional forecasting models usually adopt a global modeling approach, that is, use a unified model to cover various forecasting processes and model the relationship between independent variables and dependent variables. However, the correlation mode between variables in the data usually...

AI Technical Summary

Problems solved by technology

However, real-time forecasting is a rather complex problem, for example, for air quality, it is not only affected by recent historical air quality factors and traffic factors, but also by real-time meteorological factors

Therefore, the multi-model-based forecasting methods in the prior art, including the above-mentioned patents, are still not very effective for real-time forecasting, and the accuracy and reliability of the forecast results are still low

Images