Showing papers on "Partial autocorrelation function published in 1989"
TL;DR: In this article, the authors investigated the forecasting efficiency of an expert system, an automatic time series modeling system, when applied to a quarterly earnings per share series, and found that the intervention analysis which specifically models the outlier may enhance forecasting efficiency.
Abstract: The purpose of this study is to investigate the forecasting efficiency of an expert system, an automatic time series modeling system, when applied to a quarterly earnings per share series. The Bethlehem steel quarterly earnings series has a severe outlier problem and the intervention analysis which specifically models the outlier may enhance forecasting efficiency. The purpose of this study is to re-examine the intervention analysis of Bethlehem Steel's quarterly earnings per share series behaviour previously analyzed in Hopwood and McKeown (1986). The very large $1 billion loss of Bethlehm Steel in 1982 created an outlier in the quartely earnings series that may distort the autocorrelation and partial autocorrelation function estimates such that traditional time series models may not be appropriate [Box and Tiao (1975)]. The intervention analysis model reported in Hopwood and McKeown is substantiated in this analysis; however, the intervention modeling effort is expanded by using the intervention model i...
26 citations
TL;DR: In this paper, a limit theorem for sample partial autocorrelations was developed for the simple branching process with immigration, which was used to develop a Quenouille-type goodness-of-fit test based on sample partial auto-correlations.
23 citations
TL;DR: The authors examined the F -test for a set of linear restrictions on the parameters in the standard linear regression model and showed that the F-test is extremely non-robust to autocorrelation, in the sense that the size of the tests tends to either one or zero as correlation among disturbances increases.
12 citations
1 Jan 1989
TL;DR: In this article, an estimate of the autocorrelation coefficient based on an extreme value statistic is proposed for an AR(1) process with positive or bounded innovations, and an asymptotic essentially nonparametric confidence interval is derived.
Abstract: For an AR(1) process with positive or bounded innovations, an estimate of the autocorrelation coefficient based on an extreme value statistic is proposed. Asymptotic properties are investigated. In particular, an asymptotic essentially nonparametric confidence interval for the autocorrelation coefficient is derived.
6 citations
TL;DR: A composite linear model is proposed which generates a non-Gaussian stationary stochastic process with a given third-order autocorrelation function and a white power spectrum and results of numerical experiments confirm the validity of the model.
Abstract: . A composite linear model is proposed which generates a non-Gaussian stationary stochastic process with a given third-order autocorrelation function and a white power spectrum. The design of the model is based on the fact that a type of finite-impulse-response linear system with a non-Gaussian white input series produces an output process whose third-order correlations exist only for special time lags. An arbitrary third-order autocorrelation function can be constructed by superposing output processes of this type. The model requires at most 2L2+ 4L+ 1 independent identically distributed (i.i.d.) input processes for the third-order autocorrelation function with the largest time lag L. Results of numerical experiments confirm the validity of the model.
3 citations
[...]
TL;DR: MTS is a software package from Automatic Forecasting Systems designed to estimate vector ARIMA models on IBM PCs and PC-compatibles and as special cases it can estimate univariate ARimA models and transfer function models.
Abstract: MTS is a software package from Automatic Forecasting Systems designed to estimate vector ARIMA models on IBM PCs and PC-compatibles. As special cases it can estimate univariate ARIMA models and transfer function models. It is designed to run on IBM PCs configured with a minimum of 320K memory and two floppy drives. A math co-processor is highly recommended but not required. The program is menu-driven. The menus allow the user to carry out data manipulations (data entry, data corrections, transformations, etc.) and the Box-Jenkins identification, estimation, and forecasting steps. Like AFS's univariate AUTOBJ program, the user can identify, estimate, and forecast his own model, or allow the program to select a model automatically. The menus are easy to use. Mistakes are easily corrected by recalling the last menu to make necessary changes. Data are input into the program in separate free-format ASCII files. The program has a special utility to convert PRN or DIF files for input. Some data transformations can be carried out within MTS. Differencing the data, Box-Cox transformations (i.e. log or power transformations) are executed from menus in the identification, estimation, and forecasting modules. It is possible to carry out other scaling operations within the program, but the procedure is cumbersome. Most users will probably carry out all data transformations in a spreadsheet such as 123 or Quattro, and then write the transformed data into PRN files which can then be directly read into the program. The identification module prints out autocorrelation and partial autocorrelation matrices. Point estimates are printed next to the '+ - ' matrices advocated by Tiao and Box (1981). The joint significance of the elements in the matrices is printed, and the program offers some advice about the likely form of the model. Estimation is not carried out via maximumlikelihood. Given the large number of parameters in vector ARMA models, maximumlikelihood estimation would push most PCs over the brink. MTS uses a method of moments