scientific article; zbMATH DE number 192852

From MaRDI portal
Publication:4039746





The text and the accompanying software are suitable both for the courses of time series analysis (they were developed during the pedagogic activity of the author) and for the readers interested in the application of time series methods. Four chapters and a mathematical and statistical appendix provide in a balanced form the description of the basic theory, methods and algorithms for the time- and spectral domain analysis of univariate and bivariate time series: Chapter 1 - Describing, transforming, and forecasting univariate time series; Chapter 2 - Probability theory for univariate time series; Chapter 3 - Statistical inference for univariate time series; Chapter 4 - Analyzing bivariate time series. It includes e.g. the discussion of the traditional time series models, Box-Jenkins modeling with modern methods of identification and estimation, tests for white noise, parametric and nonparametric spectral density estimation, methods for determining periodicities, cross-spectral analysis, transfer function modeling, FFT, various aspects of regression analysis. The basic results for each major topic are usually gathered into one theorem (with proof in some cases) so that one can go directly to the material of interest. Two diskettes which are enclosed contain the TIMESLAB program. The basic requirements for the exploitation are the following: Any personal computer compatible with the IBM family of PCs (i.e. also PC/XT and PC/AT) with 640 K RAM, a numeric coprocessor, either two floppy disk drives or a hard disk, color graphics adapter or enhanced graphics adapter, and the version 2.0 or later of DOS. The program consists of approximately 150 commands each performing an operation as reading, transforming or forecasting data (they include several graphics commands). A text editor is built in, and it is possible to issue operating system commands while inside the problem. The commands can be filed to macros by the user and, moreover, 80 ready macros are provided with the program so that TIMESLAB can be used as a time series analysis programming language (the example is the macro REGAR for the regression with AR errors). The material of particular chapters is illustrated by many examples and supplemented by both theoretical and computational problems (for this purpose one uses various data sets provided with the TIMESLAB).




Cited in
(22)








This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4039746)