Re: A step too far in regression fit ?

[<pl...@pi...>]

On 10/11/11 19:44, Hans-Bernhard Bröker wrote:
> On 11.10.2011 14:04, pl...@pi... wrote:
>
>> The fit is basically making to a convergence
>
> No, it's not. It's making it to failure.
>
>> necessary to it to do the invert RtR
>
> Yes.
>
>> since it has done all I asked.
>
> No, it hasn't. It may have done all you believe you need, but not all
> you actually asked for.

What has not been done if it has already found the result and output 
that it has found a convergence? I'm not saying you're wrong but it 
would be helpful if you said what I am missing rather than suggesting 
there is something more but not saying what it is .

I asked for a regression fit. I got one , I got convergence (apparently 
according to fit) I got the results , I got the stats. I got more that I 
asked for.

What do you think I "actually asked for"?
You love being smart , being explicit would be more help.


>
>> In this case I had started with data where no extra offset was needed ,
>> ie c0=0 was correct before running fit.
>
> Then you _shouldn't_ have told fit to modify it.

I was expecting it to be slightly different from zero but close to that 
value. That made it a logical starting point . If I knew it was exactly 
zero , clearly I would not have added it as a via parameter. I may not 
quite as smart as HBB but I'm not a complete cretin.

>
> A parameter value of zero is particularly bad in that case. No fit
> parameter may be zero.
_wrong_

The following works. I frequently set zero as a starting value if if 
makes sense.

base=basedrift=0
if (use_baseline) fit  baseline(x) datafile using 1:4 via base,basedrift

>
> As it is, you pretty much explicitly asked for 'fit' to fail. No wonder
> it did.

If that were the case , I seem not to have come across that rather 
essential point it in the doc. Since you are wrong that may explain the 
absence of such information.

Gruss.