View Full Version : finding quantiles
sleck
07-31-2008, 10:42 AM
how do you find the quantile of a single number in a distribution in R?
input: number
output: percentage
THank you!
It's the p*** function, where *** is the distribution, eg pt(q, df, ncp, lower.tail = TRUE, log.p = FALSE) for a t distribution.
> pt(3, df = 9)
[1] 0.9925218
For help type ?p***
sleck
07-31-2008, 12:45 PM
the distribution I have is non normal and discontinous, I'm not sure what kind of distribution this is.
can I still use this function if I don't know what kind of distribution this is?
OK.if you have no idea about the distribution then a resampling or bootstrap strategy casn be used.
There are functions to do that in R,but you shhould get to understand what this is about
> http://cran.r-project.org/web/packages/boot/index.html
Roughly speaking, you use your sample to generate a bunch of samples that represent your true sampling distribution so as you can get a quantile by simply sorting and enumerate.
Of course you can have a quantile the way in my actuarial classes by simply typing help("quantile") on yor R console (pretty much I'm sure this will do!)
HiH
TheEcologist
07-31-2008, 01:44 PM
the distribution I have is non normal and discontinous, I'm not sure what kind of distribution this is.
can I still use this function if I don't know what kind of distribution this is?
You mean you have data (a vector of values)? or a function?
for a vector of values see:
?quantile()
In R.
TheEcologist
07-31-2008, 04:08 PM
OK.if you have no idea about the distribution then a resampling or bootstrap strategy casn be used.
There are functions to do that in R,but you shhould get to understand what this is about
> http://cran.r-project.org/web/packages/boot/index.html
Roughly speaking, you use your sample to generate a bunch of samples that represent your true sampling distribution so as you can get a quantile by simply sorting and enumerate.
Of course you can have a quantile the way in my actuarial classes by simply typing help("quantile") on yor R console (pretty much I'm sure this will do!)
HiH
Ah, your post wasn't here when I decided to post mine! :rolleyes:
sry mp83
Powered by vBulletin™ Version 4.1.3 Copyright © 2013 vBulletin Solutions, Inc. All rights reserved.