I am trying to do a quadratic regression via LINEST in Excel 2013 as described in this thread with its wonderful answer. Unfortunately it does not work for me. What I get is the following:
I am using the German version of Excel, so I have to use the function RGP which is the equivalent of LINEST. And WAHR means TRUE.
The cell in the 7th and 13th row just show the copied code from the arrays. Of course I get the same coefficients when I don't put ;;WAHR
at the end. I just want to show the entire stats to you, maybe it's important for the error diagnosis.
When I try to get a cubic regression via =RGP(B2:B5;A2:A5^{1,2,3})
Excel asks me if the formula should be corrected to =RGP(B2:B5;A2:A5^{1,23})
. If I select No, I have to change the formula, because otherwise there would be a problem.
So why do I not get 1, 2 and 1 as coefficients in the quadratic case and why does the cubic case not work at all? What could be the problem? Did anything change in Excel 2013?
I answer the question now instead of deleting it, because there may be somebody who has the same problem. And simoco obviously didn't feel like putting his comment as an answer.
As simoco already said in his comment, the separator ,
in ^{1,2}
is dependent on the localization of Excel.
Since I have the German version of Excel I have to use a dot .
instead. So everything works fine with ^{1.2}
and ^{1.2.3}
.