[R] R-help Digest

[Prof Brian Ripley]

On Fri, 6 May 2005, Sebastian Schoenherr wrote:

> Hi folks,
> I have to create my own time series, Is it possible to generate ARIMA time
> series, where i can define the range of the values in the y axis. (e.g: Values
> only between 0 and 1)

No.

Take a look at the definition of an ARIMA process.  Suppose e.g. you have 
an AR(1) process.  Then if innovations are positive and the coefficient is 
positive the value can be arbitrarily large.  You can construct all sorts 
of similar counter-examples.

This isn't a real problem is it?

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595