By Michael J. Wichura
This ebook is set the coordinate-free, or geometric, method of the idea of linear versions; extra accurately, version I ANOVA and linear regression types with nonrandom predictors in a finite-dimensional surroundings. This technique is extra insightful, extra dependent, extra direct, and easier than the extra universal matrix method of linear regression, research of variance, and research of covariance types in information. The e-book discusses the instinct at the back of and optimum houses of varied equipment of estimating and trying out hypotheses approximately unknown parameters within the types. subject matters lined comprise internal product areas, orthogonal projections, ebook orthogonal areas, Tjur experimental designs, uncomplicated distribution idea, the geometric model of the Gauss-Markov theorem, optimum and nonoptimal homes of Gauss-Markov, Bayes, and shrinkage estimators lower than the idea of normality, the optimum homes of F-tests, and the research of covariance and lacking observations.