Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!samsung!uunet!munnari.oz.au!ariel!ucsvc.ucs.unimelb.edu.au!wehi.dn.mu.oz!baxter_a
From: baxter_a@wehi.dn.mu.oz
Newsgroups: comp.sys.amiga.applications
Subject: Re: Regression Analysis
Message-ID: <1991Jun26.103503.24633@wehi.dn.mu.oz>
Date: 26 Jun 91 10:35:03 GMT
References: <57096@nigel.ee.udel.edu>
Organization: Walter & Eliza Hall Institute
Lines: 24

In article <57096@nigel.ee.udel.edu>, GORRIEDE@max.cc.uregina.ca (Dennis Robert Gorrie) writes:
> 
> I was looking for a similar program.  I tried Math-Amation, but
> I was not impressed.  So I use a spread sheet program, Analyse.
> You could use any spreadsheet program.  You just have to type
> in your data, and know the formulas for the Least Squares
> approximation, (Best Fit Line) and Pearson's product moment r.
> 
> Look them up in any stats book.  If you use a spreadsheet, then you
> can apply the formula to your all your data, get the best fit points,
> and then plot them.  Plus, a good spreadsheet program will
> be able to import/export data in LOTUS format for using your same
> data on an pc-clone.
> 
> P.S.  Physics data is not always linear, so you have to apply
> Log or Ln to your data to make it linear.  When you finally calculate
> the best fit  equation of the Log(y) vs x graph, or Ln (y) vs x, you
> then have to integrate it to get the equation of your original
> non-linear graph.

Multiplot does all this automatically. I am currently looking at improving
the curve fitting function. The latest version uses an iterative method.

Regards Alan