Advances and Applications in Statistics
Volume 45, Issue 1, Pages 59 - 74
(April 2015) http://dx.doi.org/10.17654/ADASApr2015_059_074 |
|
ACCURATE CONFIDENCE REGIONS FOR MARKOV-MODEL PARAMETERS
Jan Vrbik
|
Abstract: We demonstrate how to build accurate confidence regions for the three parameters of the usual autoregressive Markov-model, based on a random but correlated sample of n consecutive observations. We utilize the well-known fact that values of the corresponding likelihood function follow an approximate chi-square distribution, and we show how to substantially improve this approximation by including corrections inversely proportional to the sample size. |
Keywords and phrases: confidence regions, Markov-model, maximum-likelihood estimation.
|
|
Number of Downloads: 394 | Number of Views: 1120 |
|