gt;> To: PIKAL Petr
>> Cc: r-help@r-project.org
>> Subject: Re: [R] fit lognorm to cdf data
>>
>> How about
>>
>> proc <- c(0.9, 0.84, 0.5, 0.16, 0.1)
>> size <- c(0.144, 0.172, 0.272, 0.481, 0.583)
>> plot(size, proc, xlim=c(0,1), ylim=c
ject.org; PIKAL Petr ; r-help@r-
> project.org
> Subject: Re: [R] fit lognorm to cdf data
>
> * fitdistr?
>
> * it seems unusual (to me) to fit directly to the data with lognormal...
> fitting a
> normal to the log of the data seems more in keeping with the assumpti
Hi
Great. I did not think that such combination is posssible.
Thanks.
Petr
> -Original Message-
> From: peter dalgaard [mailto:pda...@gmail.com]
> Sent: Tuesday, July 11, 2017 1:11 AM
> To: PIKAL Petr
> Cc: r-help@r-project.org
> Subject: Re: [R] fit lognorm to cdf
How about
proc <- c(0.9, 0.84, 0.5, 0.16, 0.1)
size <- c(0.144, 0.172, 0.272, 0.481, 0.583)
plot(size, proc, xlim=c(0,1), ylim=c(0,1))
fit<-nls(proc~plnorm(size, log(xmid), sdlog, lower=FALSE), start=list(xmid=0.2,
sdlog=.1))
summary(fit)
lines(fitted(fit)~size)
-pd
> On 10 Jul 2017, at 16:27 ,
* fitdistr?
* it seems unusual (to me) to fit directly to the data with lognormal...
fitting a normal to the log of the data seems more in keeping with the
assumptions associated with that distribution.
--
Sent from my phone. Please excuse my brevity.
On July 10, 2017 7:27:47 AM PDT, PIKAL Pet
Dear all
I am struggling to fit data which form something like CDF by lognorm.
Here are my data:
proc <- c(0.9, 0.84, 0.5, 0.16, 0.1)
size <- c(0.144, 0.172, 0.272, 0.481, 0.583)
plot(size, proc, xlim=c(0,1), ylim=c(0,1))
fit<-nls(proc~SSfpl(size, 1, 0, xmid, scal), start=list(xmid=0.2, scal=.1)
6 matches
Mail list logo