Abstract

This paper addressed the two frequently used error diagnostics in multiple linear regression analysis namely Hadi’s influence measure and AndrewPregibon statistic which were logically extended and generalized to the multivariate form denoted as Multivariate Hadi’s influence measure 2p i Hand Multivariate Andrew-Pregibon statistic)( . The Proposed)p i AP( multivariate measures were used to identify the Potential Outliers and Influential Observations in multivariate linear regression analysis. For this, 11 multivariate regression models were fitted and the multivariate ) p i AP (2p i H, )( measures are utilized to scrutinize the residuals; results were exhibited along with the control charts.

Author: G. S. David Sam Jayakumar, A. Sulthan and W. Samuel

Received on: September, 2019

Accepted on: March, 2021