[R-sig-eco] generating equation (linear or non-linear) from scatter plot data

[Don McKenzie]

On 27-Sep-13, at 12:32 PM, shylock holmes wrote:

> Hi all - I have been looking on-line for an answer but cannot find  
> one thus far, and was hoping anyone could be so kind to assist with  
> some suggestions.
>
> I have biological data of body size (y) over time (x), and after a  
> preliminary inspection using a scatter plot the data suggest the  
> body size is declining. I used a linear regression to confirm this,  
> however, now I would like to examine what sort of underlying  
> regression equation I can use to describe this decline, i.e., how  
> do I get the parameters for the equation and for a more fitting  
> curve (better than the loess)? I could not find anything online on  
> this so far.
>
> The curve is what I call a "reversed growth curve", and I also  
> would like to examine at what year the decline is 5% or greater  
> from the horizontal/asymptote (if that makes sense)......

To get a meaningful answer you will probably have to specify more  
clearly the constraints on what you are doing.  For example, if you  
use a linear regression, *every* year will have the same slope. If I  
understand your question, the answer will depend sensitively on the  
type of curve that you fit.

This is not really an R question, and you might want to consult a  
statistician, or take a look at Ben Bolker's or Frank Harrell's books  
on regression modeling.
>
>
>
> thank you for any suggestions on this......m
>  		 	   		
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-ecology mailing list
> R-sig-ecology at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-ecology





Don McKenzie, Research Ecologist
Pacific Wildland Fire Sciences Lab
US Forest Service
phone: 206-732-7824

Affiliate Professor
School of Environmental and Forest Sciences
University of Washington