Regression / modelling help

#1
Hi All,


I'm new here so If I've posted this in the wrong place or its not practical then I apologise.


I'm new to regression so helpfully Im not asking a silly question.

I have got 3 variables of data of which one of the variables is a price ( probability ).

When I do a regression of the price against result I get a R2 number.

When I do a regression of each other variable I also get an R2 score ( near to the price score ).
When I do a regression of all 3 variables I get the best R2 score much better than any of them by themselves.

What should I do to turn all 3 variables into one probability and thus use the best regression ?
I have access to a few modelling packages and wondered whats the best way to put all 3 variables into a price and therefore a way of combining the 3 ?

Can anyone help ? or is there someone who would love to be paid for some tuition on this subject ?



Rgds
Jason
 

hlsmith

Omega Contributor
#4
If the r^2 increased that much I would wonder about effect modification. It may be of interest to investigate their interaction terms.