CYCLICALITY and SEASONALITY

What Goes Around, Comes Around

10-01-96

When we examine time series DATA, we find that many of these series have a significant underlying CYCLICAL component. Sometimes it will be SEASONAL (as in canned PORK & BEAN sales), and sometimes the periodicity will span a few years (as in WEATHER PATTERNS and NEW CAR SALES).

What this also implies is that what may seem 'Brand New' to the untrained eye, could really be a CYCLICAL REPETITION. My apologies to Santayana and Chinese Philosophy,.

05-16-07

In order to determine if a particular time series data vector actually fits a cyclical pattern, we need to fit a Sinusoidal Model to this data and see how well this model works in replicating existing data and in predicting NEW observations. This is really a case of MULTIPLE REGRESSION which requires that the data be set up for each observation as indicated in the formula above.

Once this has been done, the computational gruntwork can be handled as any other MULTIPLE REGRESSION problem.

(SEE: MULTIPLE REGRESSION)

All of this can also be handled by Module #15 TIME SERIES, in the Spring-Stat(c) statistical system.

(SEE: TURBO SPRING-STAT(c) MODULES)