Arma Model Of Arfima Model And Its Bayesian Statistical Inference And Empirical Analysis

In the economic sphere, we usually use the classic short memory ARMA time series model for an objective description of the process of economic and projections. In practice, however, because of the special nature of the economic field, we often encounter a lot of characteristics with a long memory time series. Therefore, we need to introduce time-series theory of a new system in the field of ARFIMA model. What's more, we still encounter many difficulties when using the frequency of the traditional statistical methods for economic Time series analysis. As a result, this article re-introduction of a new economic model of a Bayesian analysis for time series analysis. It provides a reasonable model framework of analyzing model framework when using Bayesian analysis of ARFIMA mode.This paper mainly studied the classical economic time series ARMA model and its simulation, and studied a long memory time series ARFIMA model and simulation. What's more, we make a research of ARFIMA model of Bayesian inference theory and its applications. Specific process is as follows:First of all, we analyze the statistical structure of the time series ARMA model, and forecast Brent oil price with the ARMA modelSecondly, analyzing the statistical structure of the ARFIMA model, then we use R / S analysis and fractional difference is derived. Through the use of Matlab, S-plus software, the Brent oil price reached a long-term memory sequences, and forecast Oil price.Finally, we give Bayesian analysis of the long memory time series ARFIMA model. Building a model of the conditions and parameters of the likelihood function of the prior distribution is derived parameters of the conditions of its posterior density and the edge of a posterior density; in the empirical analysis of the data, with WinBUGS package for the analysis of the simulation.