The Museum of HP Calculators

HP Forum Archive 21

[ Return to Index | Top of Index ]

With a Little Help From My Friends
Message #1 Posted by Namir on 5 Oct 2012, 12:47 p.m.

Hello All,

I gave a presentation at HHC2012 for Relative Error Regression. I have found only one reference on this subject (a paper by professor Chris Tofallis). Tofallis uses the title Least Squares Percentage Regression in his work. Does anyone know more about other papers or books that discuss Relative Error Regression? I would like to find more references that discuss calculating correlation coefficients and other post-regression statistics (such as confidence intervals) for the regression coefficients. Tofallis gives some basic coverage for calculating the correlation coefficient for Relative Error Regression. I would like to confirm his equations and perhaps move beyond that!

Thank you!!

Namir

PS: I am hoping to write an article for the HP Solve about relative error regression. Having as much equations on the related statistics would really help to write a good reference paper,

Edited: 5 Oct 2012, 12:50 p.m.

      
Re: With a Little Help From My Friends
Message #2 Posted by Jeff Kearns on 5 Oct 2012, 5:26 p.m.,
in response to message #1 by Namir

Namir,

I have a book that discusses (among other things) the "Significance of the Difference Between the Regression Coefficients b1 and b2 of Two Separate Equations".

Is this something you are interested in?

The book is called Statistics Manual, by Crow, Davis, and Maxfield.

Jeff

            
Re: With a Little Help From My Friends
Message #3 Posted by Namir on 5 Oct 2012, 6:24 p.m.,
in response to message #2 by Jeff Kearns

Hi Jeff,

If I am correct, the subject you mentioned compares two regression lines? Correct?

I am looking at articles that discuss relative/percentage regression error where one seeks to reduce the sum of the relative error: 

Error e(i) = y(i) - a - b*x(i) for a linear model
Relative error re(i) = (y(i) - a - b*x(i))/y(i)
                     = 1 - a/y(i) - b*x(i)/y(i)
                     = e(i)/y(i)

Namir

[ Return to Index | Top of Index ]

Go back to the main exhibit hall