Re: SPC - independent data

[CS Comms <KN3a8094@cscoms.com>]

On 7/24/03 11:15 PM, "Van Putten, Dirk [ETP/SAN]" <Dvanputten@lodanwest.com>
wrote:

> Hello DEN:
> 
> I do not understand the concept of "autocorrelation".

Simply stated autocorrelation refers to the state where data in a sequential
stream are not independent.  Each point is somehow related to the previous
point (or points) and to the subsequent points.  There are numerous
examples.  In continuous processes (e.g. Refining, paper making) changes
introduced do not happen all at once.  One does not therefore have a series
of points, an event, and then a series of subsequent points arising from
that event all at once.

The temperature in a refining column may be adjusted upward and it might
take, say 30 minutes for that change to take place.  The serial points
observed during that period are not independent, but rather share the
changing temperature conditions and thus are correlated.

I'm sure someone can do a better job explaining than that, but at least it's
a beginning...

John Dowd
jsdwd@ispwest.com