Trouble when reproducing regression lines manually

[mlc]

I have produced a simple line chart, put in a trend line and asked for the equation. The R2 is > 0,99. However, when I try to guess the values of Y, using the coefficients and a specific X, the result is incorrect. I have tried fitting a very specific curve, with 5 polynomial terms, and am seeing each of the coefficients with 20 decimals to make sure I wasnt missing anything... Has anyone else had this problem? Any idea of how to fix it?

 

[pgc01]

Hi mic
Welcome to the board

I have produced a simple line chart, put in a trend line and asked for the equation. The R2 is > 0,99. However, when I try to guess the values of Y, using the coefficients and a specific X, the result is incorrect. I have tried fitting a very specific curve, with 5 polynomial terms, and am seeing each of the coefficients with 20 decimals to make sure I wasnt missing anything... Has anyone else had this problem? Any idea of how to fix it?

No way to test anything because you did not post any data.

To get an answer you should

- post input data
- tell us which trend line you chose (for ex. 5 polynomial terms)
- the trendline equation you got, so that we can confirm it
- for some values of X the expected Y values and the Y values you get from the equation

With something like this, we can try to figure out the problem.

Note that I'm only answering because you are new to the board, or else I doubt you'd get any anwer as you did not post anything for us to work with.

Please read the "Tips for posting a question on the board" in the FAQ:

http://www.mrexcel.com/forum/faq.php

 

[shg]

It's easier (a lot easier) to get the cofficients using a LINEST formula rather than copying them from the trendline equation.

 

[roc_on_the_rocks]

The trendline has worked for me, but only when I increase the number of decimal points to 10 or 20, depending on the equation.

LINEST? Uh, I need to look into that!