Re: [Gnuplot-info] Gnuplot "fit" command

[<br...@ph...>]

ra...@be... wrote:
> Perhaps the mailing list does not allow attachements. 

I usually does... last I checked, there was only a size restriction, but 
no general refusal of attachments.

> 1,0.4

That's your problem right there.  Comma-separated isn't gnuplot's native 
input format.  You have to change the format, or tell gnuplot about it.

> It seems unlikely that playing with scaling factors should be necessary as
> this is the simplest data set possible and both variables are in the same
> order of magnitude. 

The scale of x is completely irrelevant anyway --- that of y only 
matters if you failed to provide realistic y errors.

Difference of scales matters if you have more than one fitted parameter, 
and they're of vastly different magnitude.  Scaling of x and y only 
matters as far as it causes parameter scales to differ --- but that can 
always be fixed in the model.