Error estimates with r2 in linear regression

[mh8782]

I have been given the results of some regression analysis, where some variable y is a function of x1 and x2. I have a r2 value for the analysis, and I would like to know if I can use that info to estimate a maximum and minimum value for y (say some 95% estimates), given x1 and x2. e.g. y could be as high as 20, and as a low as 10.

Is that possible, and if so, how could I do it?

 

[vinux]

What I understand from your question is:
can we come up with CI for y ( given x1 and x2) by only using Multiple R2 ?

No. It is not possible.

If you have some assumptions/ or information about regression coefficient then some attempt can be done.

 

[mh8782]

What if one were to assume that the residuals were distributed normally from the line of best fit? Would that change things?

Thanks for your reply