Sign-based Methods in Linear Modelsкнига

Работа с книгой


[1] Boldin M. V., Simonova G. I., Tyurin Y. N. Sign-based Methods in Linear Models. — United States: United States, 1997. — 236 p. In the book a new nonparametric approach to the analysis of statistical data is exposed. It consists in using only the signs of observations or of certain functions of them depending on the data structure. Hence the approach is referred to as sign based. The book treats regression and autoregression models important for applications. For these models the sign-based methods yield the solutions of the principal statistical problems (parameter estimation, hypothesis testing, etc.). Both finite-sample and large-sample properties of the sign procedures are studied. The sign procedures are shown to be robust with respect to gross errors in the data. Numerical algorithms to implement the sign analysis are proposed, and examples of their application to real and simulated data are given. The exposition evolves from elementary to advanced theory to make the book accessible to a broader readership. The book is intended for those studying or applying mathematical startistics.

Публикация в формате сохранить в файл сохранить в файл сохранить в файл сохранить в файл сохранить в файл сохранить в файл скрыть