TY - JOUR
T1 - Prediction via estimating functions
AU - Thavaneswaran, A.
AU - Heyde, C. C.
PY - 1999/2/15
Y1 - 1999/2/15
N2 - This paper is concerned with prediction methods for linear as well as non-linear non-Gaussian models. Recursive formulas are obtained by combining the information associated with the predictive functions. Nonlinear predictors are obtained for linear and nonlinear time series models. The innovation algorithm is shown to be a special case of the proposed algorithm for linear processes with known autocovariances. Least absolute deviation predictors are shown to be a special case as well. Recursive prediction incorporating the variability due to parameter estimation is also discussed in some detail.
AB - This paper is concerned with prediction methods for linear as well as non-linear non-Gaussian models. Recursive formulas are obtained by combining the information associated with the predictive functions. Nonlinear predictors are obtained for linear and nonlinear time series models. The innovation algorithm is shown to be a special case of the proposed algorithm for linear processes with known autocovariances. Least absolute deviation predictors are shown to be a special case as well. Recursive prediction incorporating the variability due to parameter estimation is also discussed in some detail.
KW - Estimating functions
KW - Least absolute deviation
KW - Minimum mean squares
KW - Recursive prediction
KW - Stable distributions
UR - http://www.scopus.com/inward/record.url?scp=0005723510&partnerID=8YFLogxK
U2 - 10.1016/S0378-3758(98)00179-7
DO - 10.1016/S0378-3758(98)00179-7
M3 - Article
SN - 0378-3758
VL - 77
SP - 89
EP - 101
JO - Journal of Statistical Planning and Inference
JF - Journal of Statistical Planning and Inference
IS - 1
ER -