Time series analysis has many applications. How it is applied, depends very much on the particular application. The following information is intended as a guide to sales, marketing and economics applications. Time domain time series models are limited in their ability to represent cycles in the data. The moving window spectral antithetic (MWSA) time series model is a frequency domain approach. It is an extension of the time series concept, to a generalized automatic system that decomposes the data into trend, periodic components and residuals. FOURCAST employs the MWSA Method and is particulary effective when the time series contains cyclical components. It uses spectral analysis to perform an in depth analysis of the trend and cyclical components. A moving time domain window is used to determine how each cyclical component is changing over time. One component may be growing while another component may be shrinking (no other method can do this). The cycles are forecast individually, then recombined. The forecast is dynamic. When the forecast cycles are in alignment, turning points and changes in trend will occur. Also, antithetic forecasting (combining 2 forecasts with negatively correlated errors) is used to reduce bias and minimize forecasting error. The main objective of this information, and the test that follows is a critical examination of the the way in which we view this application, and in general, the role of time series analysis and forecasting.
ASSUMPTION: Sales, marketing & economic data are made up of trend, cyclical, and random components. Plot a time chart of the data. Review the data for trend, cycles and changing variance. Plot a histogram. Review the histogram for normality or lognormality. Consider what transformations may be used to make the data stationary. Consider what window length would match the longest dominant cycle. Select the appropriate transformations and window length then estimate the model.
Questions
Answer each of the following multiple choice review questions by
selecting the response which is most immediately appropriate.
FOURCAST Home Page
http://www.fourcast.net