Improved combined prediction method based on PSO-RBF hybrid correction operator

Abstract

The invention belongs to the technical field of price prediction, and particularly discloses an improved price combination prediction method based on a PSO-RBF hybrid correction operator. The method comprises the following steps of determining a research object, and collecting price time series data of the research object; constructing a plurality of price basic prediction models for small sampleprediction and a plurality of price basic prediction models for large sample prediction; taking the fitting sample data set as an input, taking the predicted price as an output, based on a radial basis function, adopting a particle swarm algorithm to carry out basic model improvement and combined model improvement on the price basic prediction model, and taking MSE, MAPE and Tel coefficient criteria as combined prediction evaluation criteria to construct a target function of the price combined prediction model; constructing a price combination prediction model according to the weight of the price basic prediction model; and verifying the prediction effect of the price combination prediction model. According to the invention, the problem of fusion of long-term memory and short-term law research of time series data in a complex economic environment is solved.

Description

technical field [0001] The invention belongs to the technical field of price forecasting, and more specifically relates to an improved combined forecasting method based on a PSO-RBF hybrid correction operator. Background technique [0002] The emergence of combined forecasting theory is a major change to the traditional forecasting theory, which greatly enriches the theoretical system of traditional forecasting. At present, the combination forecasting theory provides a new idea and method for the forecasting research of economic data under the characteristics of time series due to its wide application range, strong robustness, and high forecasting accuracy. However, considering the non-linearity, randomness, heteroscedasticity and other characteristics of the superimposed effect of external noise on economic data, especially for price forecasting under the complex economic system environment and high precision requirements, traditional combination forecasting The conclusion...

AI Technical Summary

Problems solved by technology

[0005] In view of the above defects or improvement needs of the prior art, the present invention provides an improved price combination forecasting method based on the PSO-RBF hybrid correction operator, by constructing a plurality of small-sample forecasted basic price forecast models and large-sample forecasted price Based on the basic forecasting model, improve the basic model and the combination model to build a price combination forecasting model, complement each other through large-sample forecasting and small-sample forecasting models, and make up for the natural deviation of the price sample size selection of the research object for forecasting results. It solves the problem of integration of long-term memory and short-term law research of time series data in a complex economic environment

Images