1999 年 29 巻 1 号 p. 23-38
In a generalized linear regression model with autocorrelated error term, some shrinkage type regression predictors are examined in a situation where the prediction area is different from the sample area. The goodness of the shrinkage regression predictor is appraised on the basis of the expectations of prediction Mean Square Error(PMSE)with respect to the model condition parameter having a known prior distribution. A modified shrinkage type predictor is proposed. Moreover, simulation results are presented to exhibit the efficiency of the proposed predictor.