Research On Modeling And Prediction Based On Multi-Variable Granular Time Series

Time series prediction research is an important content in the field of data analysis.By integrating the data information of things and analyzing the time series,we can explore the development trend and development law of things and forecast the change of data.Multivariate fuzzy time series model has its unique advantages in predicting the semantic value sequence after data fuzzification.By combining the particle calculation with the multi-variable fuzzy time series model,the division process is optimized,and the prediction effect of the model is further optimized by constructing the combined model.Firstly,the multi-variable fuzzy time series is modeled on the basis of historical data and its related factors,and the construction process of the model is optimized.In this paper,a method of constructing information particles is proposed based on fuzzy sets.In the stage of domain division,the clustering algorithm is used to divide the universe into several subintervals,and the interval length is adjusted by information granularity.In the stage of fuzzy rule reasoning,the classification ability of support vector machine model is used to realize the prediction of fuzzy sets.Secondly,In order to reduce the systematic deviation of the single model in modeling,a time series combinatorial model with adaptive variable weight is presented.The model weight is adjusted adaptively according to the prediction error of each singal model at historical time.A combination of improved multivariate fuzzy time series model and function coefficient autoregressive model is proposed to construct a combined model.The granular calculation method is used to determine the initial value of the autoregressive model parameters of the function coefficients and the adaptive multi-population genetic algorithm is used to optimize the model parameters.Finally,the proposed model is applied to the Taiwan stock index,Taipei city temperature data,hemodialysis data of kidney disease,and the prediction effect is compared with the existing methods to verify the effectiveness of the model.It can be seen from the simulation experiment that the model used in this paper can effectively improve the prediction accuracy.