When you fit the model the first time, you subtracted some value from
each data-point (centered the data) and divided some value into each
data-point (scaled the data). Often the amount subtracted is the
mean of a dataset, and the amount divided by is the standard
deviation. But it doesn't have to be.
If you now write out the equation for the fitted model, you will use
x-prime or y-prime, where the -prime values are transformed. You can
call this 'transformed space' for the -prime values and 'real space'
for the values as you recorded them.
Your model applies for the transformed space - nothing else.
To check how well new data falls into your model, you must transform
them _in exactly the same way_. Subtract the same value, and divide
by the same value. Now you have the new data as -prime new data, and
your model applies to it also.
And now you can see how well your new data-prime fits the predicted -
prime values.
Cheers,
Jay
On Jan 21, 2009, at 3:39 PM, Angelica Neisa wrote:
> Dear all stat
>
> I fit a model and in the process I centered and scaled the
> variables. Now I want to use the model with other data, should I
> center and scaled the variables? if yes, what mean should I use the
> original one or the mean of the new data?
>
> thanks a lot,
> Angelica
>
> _________________________________________________________________
> Windows Live™: E-mail. Chat. Share. Get more ways to connect.
> http://windowslive.com/explore?
> ocid=TXT_TAGLM_WL_t2_allup_explore_012009
Jay Warner
Principal Scientist
Warner Consulting, Inc.
4444 North Green Bay Road
Racine, WI 53404-1216
USA
Ph: 262.634.9100
FAX: 262.681.1133
email: [log in to unmask]
Temporary: [log in to unmask]
web: www.a2q.com
The A2Q Method (tm) -- What do you want to improve today?
|