A0962
Title: Fast Bayesian estimation of dynamic linear regression models for semi long memory time series
Authors: Thomas Goodwin - University of Technology, Sydney (Australia)
Matias Quiroz - University of Technology Sydney (Australia) [presenting]
Robert Kohn - University of New South Wales (Australia)
Abstract: Dynamic linear regression models forecast the values of a time series based on a linear combination of a set of exogenous time series while incorporating a time series process for the error term. This error process is often assumed to follow an autoregressive integrated moving average (ARIMA) model, or seasonal variants thereof, which are unable to capture a long-range dependency structure of the error process. A novel dynamic linear regression model that incorporates the long-range dependency feature of the errors is proposed, showing that it improves the model's forecasting ability. A Markov chain Monte Carlo method is developed to fit general dynamic linear regression models based on a frequency domain approach that enables fast approximate Bayesian inference for large datasets. It is demonstrated that the approximate algorithm is much faster than the traditional time domain approach based on the Kalman filter while retaining high accuracy.