Regression Diagnostics for Autocorrelated Models with Moving Average Errors

Abstract

ABSTRACT:

Autocorrelation in a regression model is a common phenomenon in most practical studies. The presence of autocorrelated errors require the use of the generalized least-squares technique in estimating the regression coefficients. Using the deletion technique in checking for outliers in such a model leads to the disruption of the error structure. In this paper we take account of this disruption in studying the diagnostics of a regression model whose errors follow a first-order moving average process.