ARMA model identification. (English) Zbl 0754.62071
Springer Series in Statistics. Probability and its Applications. New York: Springer-Verlag. xi, 200 p. (1992).
The book gives an account of identification methods for univariate ARMA models. The author describes the autocorrelation methods ( Box and Jenkins method, the inverse autocorrelation method), penalty function methods ( Akaike’s FPE and AIC criteria, Parzen’s method, Bayesian information criterion, Hannan and Quinn’s criterion), innovation regression methods (Hannan and Rissanen’s method, Koreisha and Pukkila’s method) and pattern identification methods (\(R\) and \(S\) arrays, the corner method, the GPAC, ESACF, SCAN and Woodside’s methods). The last chapter is devoted to testing hypothesis methods (e.g. based on the portmanteau statistic).
The author presents fundamental ideas, historical notes, a few personal views of leading statisticians, and many references (749 items on 48 pages). No numerical examples and no methods for identification of multivariate models are given.
The author presents fundamental ideas, historical notes, a few personal views of leading statisticians, and many references (749 items on 48 pages). No numerical examples and no methods for identification of multivariate models are given.
Reviewer: J.Anděl (Praha)
MSC:
62M10 | Time series, auto-correlation, regression, etc. in statistics (GARCH) |
62-02 | Research exposition (monographs, survey articles) pertaining to statistics |
62-01 | Introductory exposition (textbooks, tutorial papers, etc.) pertaining to statistics |