Bayesian Estimation of Gold Index's Change Point of Time Series Model

摘要

The thesis mainly estimates the change point of time series model through Bayesian method. First, through establishing the time series model, adopting conjugate prior distribution, linear regression and prior information, the parameter values related to distribution can be got. Then posterior distribution and change point can be got through computation. To reduce iterative error, Peak Algorithm is used to check the posterior distribution. Finally, the gold index's change point of time series model can be got through this method.