Optimal directional statistic for general regression

Publication date: Available online 7 August 2018Source: Statistics & Probability LettersAuthor(s): Jonathan Gillard, Anatoly ZhigljavskyAbstractFor a general linear regression model we construct a directional statistic which maximizes the probability that the scalar product between the vector of unknown parameters and any linear estimator is positive. Special emphasis is given to comparison of this directional statistic with the BLUE and explaining why the BLUE could be relatively poor. We illustrate our results on analytical and numerical examples.

Source: Statistics and Probability Letters - August 8, 2018 Category: Statistics Source Type: research