# [SOLVED] Mean or Median Values?!?!

• Feb 16th 2008, 11:54 AM
frozenflames
[SOLVED] Mean or Median Values?!?!
cccc
• Feb 16th 2008, 12:55 PM
CaptainBlack
Quote:

Originally Posted by frozenflames
The following is regression data to predict customer spending for a catalog company...

Based on the data below....the amount spent is the response variable while the others (ie: closeness to other stores stores nearby, salary, children, catalogs) are predictor variables

http://img403.imageshack.us/img403/6875/databs7.png

My question is: how do I decide whether the prediction of Amount Spent is based on the mean or median values of the predictor variables in the final model?? I'm sure that I have to determine for each variable, whether the mean or median should be used and am a little confused about which way to go??? ANY HELP WOULD BE APPRECIATED!!!!!!!

Is it not the case that the regression model is build on the actual data and not the summary statistics?

RonL
• Feb 16th 2008, 07:42 PM
CaptainBlack
Quote:

Originally Posted by frozenflames
Yes, the regression model is based on the data.

But, the attachment above is indicative of my results in the context of the case. In other words, it's a prediction of AmountSpent.

I am trying to determine which value (average or the median) is a better option to predict AmountSpent. and on what factors is a particular choice the better option?

The regression model gives you anount spent as a function of the
independent variables. In any particular case you use the particular values of
the independent variables.

Assuming a linear model the prediction of the mean amout spent for a
collection of cases is the prediction of the model using the means of the
independent variables for the collection.

RonL