multi-variables Regression

[abalkees]

thread770-214997

Hi everyone,
I am trying to find the regression equation for a set of data with 2 independent variables using LINEST function in EXCEL.

I re-tabulated my data in order to use the LINEST function. The equation I would like to get is from the second order like this:
Y = c + a1*x1 + a2*x1^2 + b1*x2 + b2*x2^2

and like this as well:
Y = c + a1*x1 + a2*x1^2 + b1*x2 + b2*x2^2 + b3*x1*x2

I tried to use the same method used in this link but I could not get any results.

http://www.eng-tips.com/viewthread.cfm?qid=214997#post

I have attached my EXCEL file, have a look please.

http://files.engineering.com/getfil...-b3eb54c58fa2&file=Scaling_-using_LINEST.xlsx

May any one please help me with that?
Please give some explanation about how to get the coefficients such as the ones included in the referred thread =LINEST(Y1:Y10,B1:B10^{1,2,3,0,0,0,1,2,3}*C1:C10^{0,0,0,1,2,3,1,1,1},TRUE,TRUE)


I would like to mention something else: my original data has 4 variants Y=f(x1,x2,x3,x4) but I tried to make my question simple. Therefore, should I use the same method when dealing with more complex data?

Many Thanks
Alaa

 

[LiteYear]

The ^{0} trick wont work because you have zeroes in your data and 0^0 is undefined. Instead, just create the x1*x1, x2*x2 and x1*x2 columns and refer to them explicitly:

=LINEST(D3123,E3:H123,TRUE,FALSE)

or E3:I123 to include the x1*x2 term.

Remember to hit ctrl-shift-enter when entering the formula.

 

[DRWeig]

See attached if you have Excel 2007 or later.

I use the regressions in the Analysis Toolpack. Much more thorough, and lots of options.





Best to you,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies