# Re: [R] Question concerning pdfs

From: Greg Snow <Greg.Snow_at_imail.org>
Date: Wed, 16 Jan 2008 13:27:04 -0700

You may want to look at the logspline package. It is another way of estimating densities and allows you to specify boundries on the domain of x.

```--
Gregory (Greg) L. Snow Ph.D.
Statistical Data Center
Intermountain Healthcare
greg.snow_at_imail.org
(801) 408-8111

> -----Original Message-----

> From: r-help-bounces_at_r-project.org
> [mailto:r-help-bounces_at_r-project.org] On Behalf Of Rainer M Krug
> Sent: Wednesday, January 16, 2008 2:35 AM
> To: r-help
> Subject: [R] Question concerning pdfs
>
> Hi
>
> I have a dataset of absolute growth rates g ranging from
> close to 0 to around whatever of which I want to calculate a pdf.
> If I use density(g) the pdf will extend to below 0 so I
> logtransform g and do d <- density(log(g)).
> Now I would like to transform this pdf back, i.e. d\$x <-
> exp(d\$x) but what do I have to do with d\$y?
>
> (below is a small example)
>
>
> Rainer
>
>
> g <- exp(rnorm(100)) ## Just to generate the example data
>
> d <- density(log(g))
> dx <- d\$x
> dp <- d\$y
>
> sum(c(0, (diff(dx))) * dp) ## this is equal to one plot(dx, dp)
>
> dx2 <- exp(dx)
> dp2 <- ???????          ## what should I do here?
>
> plot(dx2, dp2)          ## what should I do here?
> sum(c(0, (diff(exp(dx2)))) * dp2)    ## this should be one
>
>
> sum( c(0, (diff(dx2))) * ( dp1 / c(1, (diff(exp(dx2)))) * c(0,
> (diff(dx1)))) )               ##This is obviously also one, but can I
> use this to define my dp2? i.e.
>
> dp2 <- dp1 / c(1, (diff(exp(dx2)))) * c(0, (diff(dx1)))) ###????
>
> ______________________________________________
> R-help_at_r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

______________________________________________
R-help_at_r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help