Skip to main content
main-content

Über dieses Buch

The analysis of time series data is an important aspect of data analysis across a wide range of disciplines, including statistics, mathematics, business, engineering, and the natural and social sciences. This package provides both an introduction to time series analysis and an easy-to-use version of a well-known time series computing package called Interactive Time Series Modelling. The programs in the package are intended as a supplement to the text Time Series: Theory and Methods, 2nd edition, also by Peter J. Brockwell and Richard A. Davis. Many researchers and professionals will appreciate this straightforward approach enabling them to run desk-top analyses of their time series data. Amongst the many facilities available are tools for: ARIMA modelling, smoothing, spectral estimation, multivariate autoregressive modelling, transfer-function modelling, forecasting, and long-memory modelling. This version is designed to run under Microsoft Windows 3.1 or later. It comes with two diskettes: one suitable for less powerful machines (IBM PC 286 or later with 540K available RAM and 1.1 MB of hard disk space) and one for more powerful machines (IBM PC 386 or later with 8MB of RAM and 2.6 MB of hard disk space available).

Inhaltsverzeichnis

Frontmatter

1. Introduction

Abstract
The time series programs described in this manual are all included in the package ITSM (Interactive Time Series Modelling) designed to accompany the book Time Series: Theory and Methods by Peter Brockwell and Richard Davis, (Springer-Verlag, Second Edition, 1991). With this manual you will find two versions of the package, ITSM41 and ITSM50 (each on a 3 1/2″ diskette). The system requirements for ITSM41 are fewer than for ITSM50 (see Section 1.2), however ITSM50 can handle larger data sets (univariate series with up to 20000 observations and multivariate series with up to 10000 observations of each of 11 components). Both versions of the package contain the programs listed below.
Peter J. Brockwell, Richard A. Davis

2. PEST

Abstract
Double click on the icon labelled pest in the itsmw window (or in DOS type PEST↩ from the C:\ITSMW directory) and you should see the figure displayed in Figure 2.1. Then press any key and you will see the Main Menu of PEST as shown in Figure 2.2.
Peter J. Brockwell, Richard A. Davis

3. SMOOTH

Abstract
To run the program SMOOTH, double click on the icon labelled smooth from the itsmw window (or in DOS type SMOOTH from the C:\ITSMW directory). After pressing ↩ to clear the screen of the title page, you will be asked if you wish to [Enter data] or to [Exit SMOOTH]. Type E and select the name of the data file to be smoothed. After following the program prompts, you will see the Smoothing Menu which provides a choice of three smoothing methods for the series {Xt, t = 1,…, n}.
Peter J. Brockwell, Richard A. Davis

4. SPEC

Abstract
To run the program SPEC, double click on the icon labelled spec in the itsm window (or in DOS type SPEC↩ from the C:\ITSMW directory). After pressing ↩ to clear the title page, you will see a menu offering the choice of spectral analysis for one or two data sets. For spectral analysis of a univariate series, type O and select the name of the data file to be analyzed. Once the data has been read in, a menu will appear which is practically identical to the Spectral Analysis Menu of PEST(in the option [Non para metric spectral estimation (SPEC)] of the Main Menu). For instructions on the use of the univariate options available in this menu see Section 2.7.
Peter J. Brockwell, Richard A. Davis

5. TRANS

Abstract
To run the program TRANS, double click on the icon labelled trans in the itsm window (or in DOS type TRANS↩ from the C: \ITSMW directory). After pressing ↩ to clear the title page, you will see the Main Menu (Figure 5.1) with five options and two explanatory statements.
Peter J. Brockwell, Richard A. Davis

6. ARVEC

Abstract
The program ARVEC fits a multivariate autoregression of any specified order p < 21 to a multivariate time series {Yt = (Y t 1,…, Y tm )′, t = 1,…, n}. To run the program, double click on the icon arvec in the itsmw window (or type ARVEC ↩ from the DOS prompt) and you will see a title page followed by a brief introductory statement describing the program. After reading this statement, follow the program prompts, selecting the option [Enter data] by typing the highlighted letter E. You will then be asked to enter the dimension m ≤ 6 (m ≤ 11 for ITSM50) of Yt and to select the file containing the observations {Yt,t = 1,…,n}. For example, to model the bivariate data set LS2.DAT you would enter the dimension m = 2 and then select the file LS2.DAT from the list of data files. The data must be stored as an ASCII file such that row t contains the m components, Y t = {Y t 1,…,Y tm )′, each separated by at least one blank space. (The sample size n can be at most 700 for ITSM41 and 10000 for ITSM50.) The value of n will then be printed on the screen and you will be given the option of plotting the component series.
Peter J. Brockwell, Richard A. Davis

7. BURG

Abstract
Like ARVEC, the program BURG fits a multivariate autoregression (of order p < 21) to a multivariate time series {Y t = (Y t1 ,…, Y tm )′ t = 1,…, n}. To run the program, double click on the icon burg in the itsmw window (or type BURG ↩ from the DOS prompt) and you will see a title page followed by a brief introductory statement describing the program. After reading this statement, follow the program prompts, selecting the option [Enter data] by typing the highlighted letter E. You will then be asked to enter the dimension m ≤ 6 (m ≤ 11 for ITSM50) of Y t and to select the file containing the observations {Y t ,t = 1,…,n}. For example, to model the bivariate data set LS2.DAT you would enter the dimension m = 2 and then select the file LS2.DAT from the list of data files. The data must be stored as an ASCII file such that row t contains the m components, Y t = (Y t 1,…,Y tm )′, each separated by at least one blank space. (The sample size n can be at most 700 for ITSM41 and 10000 for ITSM50.) The value of n will then be printed on the screen and you will be given the option of plotting the component series.
Peter J. Brockwell, Richard A. Davis

8. ARAR

Abstract
To run the program ARAR, double click on the arar icon in the itsmw window (or type ARAR↩ from the DOS prompt) and press ↩. You will then see a brief introductory statement. The program is an adaptation of the ARARMA forecasting scheme of Newton and Parzen (see The Accuracy of Major Forecasting Procedures, ed. Makridakis et al., John Wiley, 1984, pp.267 – 287). The latter was found to perform extremely well in the forecasting competition of Makridakis, the results of which are described in the book. The ARARMA scheme has a further advantage over most standard forecasting techniques in being more readily automated.
Peter J. Brockwell, Richard A. Davis

9. LONGMEM

Abstract
The program LONGMEM is designed for simulation, model-fitting and prediction with ARIMA(p, d, q) processes, where −.5 < d <.5.
Peter J. Brockwell, Richard A. Davis

Backmatter

Weitere Informationen