Re: [Gnuplot-info] fiting exponential does not match

[terriblement <ric...@gm...>]

Thank you !

I tried what you said, and the curve had a too big offset, so I've put
c=100.
Now it fits good although the values between 25 and 65 are not perfect; I
think there is nothing to do.

(f(x) = 14.2169 * exp(0.0486079*x) + 100) 

(sorry if I don't respect the Quote rules, I haven't understood them I
think)



Thomas Sefzick wrote:
> 
> your fit parameters are not independent:
> 
> 'c', which is used to compensate an 'offset', depends on 'a',
> because
> a*exp(0) = a*1
> which means that 'c+a' should be '116'.
> 
> you could set 'c' to 116, 'a' and 'b' to some reasonable
> value (e.g. a=1, b=0.02) and fit 'via a,b', adjust 'c' and
> fit again.
> 
> 
> terriblement wrote:
>> 
>> Hi everybody,
>> 
>> I have a txt file with some values, and I want to fit a curve with my
>> values...(see attached file)
>> 
>>  http://old.nabble.com/file/p32115817/test.txt test.txt 
>> 
>> it looks like an exponential function, so I'Ve tried this :
>> 
>> f(x) = a* exp (b*x) + c
>> c = 116
>> fit f(x) "test.txt" using 1:2 via a,b,c
>> 
>> and the values are totally wrong :s
>> I don't understand what I'm doing wrong, if somebody could help me (I'm
>> noob with gnuplot)
>> 
>> thanks a lot.
>> 
> 
> 
:clap:
-- 
View this message in context: http://old.nabble.com/fiting-exponential-does-not-match-tp32115817p32129436.html
Sent from the Gnuplot - User mailing list archive at Nabble.com.