Resampling for order estimation of autoregressive models with missing data

Abstract

In this artticle, we consider the order estimation of autoregressive modelswith incomplete data using expectation maximization(EM) algorithm based information criteria.The criteria take the form of a penalization of the conditionnal expectation of the log-likelihood. The ealuation of the penalization term generally involves numerical differenciation and matrix inversion. We introduce a simplification of the penalization term for autoregressive model selection and we propose a penalty factor based on a resampling procedure in the criteria formula. The simulation results show the improvement yielded by the proposed method when compares to the classical information criteria for model selection with incomplete data.