Linear Regression Model With Inequality Constraints, Statistical Diagnostics

In this paper, we study the statistical diagnostics for the inequality constrained regression problems, through case deleted and mean shift we can get the outliers, and we also prove the equality for the two ways. Then we study the residual analysis for inequality-constrained, for the model of variance inflation we study the score test and the test of homogeneity of variance, and then we discuss the likelihood estimate. At last, through influence analysis for inequality-constrained regression problems, we can get some test statistics, for example cook statistic and welsch-kuh statistic, likelihood displacement, best linear unbiased estimate and so on. In addition, a simulate data example is provided for illustration purpose.