From help-octave-request at bevo dot che dot wisc dot edu  Tue Dec 15 13:28:48 1998
Subject: Re: fit data - trendline through zero
From: heberf <heberf at calvin dot wustl dot edu>
To: help-octave at bevo dot che dot wisc dot edu
Date: Tue, 15 Dec 1998 13:29:34 -0600 (CST)


Polyfit just does a regression of Y on a constant, X, X^2, X^3 and so on up to 
the order of the polynomial you are interested in.  To set the intercept to zero 
you need to create some new variables.

	y = Y - mean(Y);
	x = X - mean(X);
	x2 = X.^2 - mean(X.^2);
etc

Put the x, x2, x3 etc into a matrix Z.  The values of the line you want will be

	f = Z*inv(Z'*Z)*Z'y;

Now a simple

	plot(x,f)

will give the plot you want.  This is a true least squares fit.  In fact the 
slope coefficient on xn will be the same as the slope coefficient on X^n in the 
original data.  The only thing that has changed is that the intercept is zero.

Heber Farnsworth.


> I use polyfit to find the trendline in my data. How can I force it to go
> through zero ?
> 
> Thanks
> 
>   Marcel
> 
> 
> ---------------------------------------------------------------------------
> 
> Marcel Meissel
> 
> Polymer Processing Laboratory
> Department of Mechanical Engineering
> University of Maryland
> College Park, MD 20742
> 
> (301)-405-2119
> ---------------------------------------------------------------------------
> 
> Visit  "www.MEISSEL.com"  and tell me about the design. I need Feedback
> !
>          _    _
>         / /  (_)__  __ ____  __
>        / /__/ / _ \/ // /\ \/ /  ... The choice of a
>       /____/_/_//_/\_,_/ /_/\_\         GNU generation
> 


------------- End Forwarded Message -------------