Conditional Predictive Ordinate (CPO) plots for ARMA model selection

Conditional Predictive Ordinate (CPO) plots for ARMA model selection

0.00 Avg rating0 Votes
Article ID: iaor200934
Country: Germany
Volume: 12
Issue: 2
Start Page Number: 211
End Page Number: 219
Publication Date: Jun 2004
Journal: Central European Journal of Operations Research
Authors: ,
Keywords: time series & forecasting methods
Abstract:

Autoregressive moving average (ARMA) models with different distributional assumptions can be compared by conditional predictive ordinate (CPO) plots if exact likelihood evaluations are not feasible. In our application we show how hyperbolic and t-distributed ARMA estimates can be compared for different model orders, when the estimation method is based on Bayes MCMC estimates.

Reviews

Required fields are marked *. Your email address will not be published.