|
|||||
|
|
Non-Gaussian series and series with non-zero means: Practical implications for time series analysis |
|
| Authors: | Eward J. Lusk Haviland Wright |
| Affiliation: | Department of Social Systems Sciences, The Wharton School, University of Pennsylvania, Vance Hall CS, Philadephia, PA 19104, U.S.A.;Department of Statistics and Operations Research, The University of Denver, Denver, CO 80208, U.S.A. |
| Abstract: | This paper presents a discussion of the relationship between the distribution of the input and the distribution of the output of linear filters. The purpose of the discussion is to focus attention on several fundamental aspects of developing ARIMA models which seem to be overlooked routinely in practice. The authors discuss the implications of non-normally distributed time series realizations and examine the effect of failure to estimate the constant parameter in developing ARIMA models. |
| Keywords: | Linear filters ARIMA models distribution of outputs of linear filters practical implications of linear filtering for developing ARIMA models |
| 本文献已被 ScienceDirect 等数据库收录！ | |