Premium
Cross‐validation Criteria for Setar Model Selection
Author(s) -
De Gooijer Jan G.
Publication year - 2001
Publication title -
journal of time series analysis
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.576
H-Index - 54
eISSN - 1467-9892
pISSN - 0143-9782
DOI - 10.1111/1467-9892.00223
Subject(s) - setar , mathematics , model selection , akaike information criterion , autoregressive model , cross validation , statistics , selection (genetic algorithm) , monte carlo method , information criteria , econometrics , autoregressive integrated moving average , computer science , star model , artificial intelligence , time series
Three cross‐validation criteria, denoted by respectively C , C c , and C u , are proposed for selecting the orders of a self‐exciting threshold autoregressive (SETAR) model when both the delay and the threshold value are unknown. The derivation of C is within a natural cross‐validation framework. The criterion C c is similar in spirit as AIC c , a bias‐corrected version of AIC for SETAR model selection introduced by Wong and Li (1998). The criterion C u is a variant of C c having a similar poperty as AIC u , a model selection proposed by McQuarrie et al . (1997) for linear models. In a Monte Carlo study, the performance of each of the criteria C , C c , C u , AIC, AIC c , AIC u , and BIC is investigated in detail for various models and various sample sizes. It will be shown that C u consistently outperforms all other criteria when the sample size is moderate to large.