Imputation of Missing Values for BL (P,0,P,P) Models with Normally Distributed Innovations

摘要: This study derived estimates of missing values for bilinear time series models BL (p, 0, p, p) with normally distributed innovations by minimizing the h-steps-ahead dispersion error. For comparison purposes, value based on artificial neural network (ANN) and exponential smoothing (EXP) techniques were also obtained. Simulated data was used in study. 100 samples size 500 each generated (1, 1, 1) using R-statistical software. In sample, observations created at positions 48, 293 496 estimated these methods. The performance criteria to ascertain efficiency mean absolute deviation (MAD) squared error (MSE). found that optimal linear most efficient estimating p). recommends OLE innovations.