[R] generate variable y to produce excess zero in ZIP analysis

Jagat.K.Sheth at wellsfargo.com Jagat.K.Sheth at wellsfargo.com
Thu Feb 21 20:06:31 CET 2013


Also something like 

## generate zero-inflated Poisson data
n <- 100
x1 <- rnorm(n)
x2 <- rnorm(n)
ind <- rbinom(n,1,1/(1+exp(-1-0.1*x1)))
y <- ifelse(ind,rpois(n,lambda=exp(1+0.2*x2)),0)

There is at least zeroinfl in the CRAN package pscl and fmr in Lindsey's non-CRAN package gnlm for fitting zero-inflated regression models via maximum likelihood.

> -----Original Message-----
> From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-
> project.org] On Behalf Of S Ellison
> Sent: Wednesday, February 20, 2013 8:03 AM
> To: lili puspita rahayu; r-help at R-project.org
> Subject: Re: [R] generate variable y to produce excess zero in ZIP
> analysis
> 
> > Subject: [R] generate variable y to produce excess zero in
> > ZIP analysis
> 
> Maybe something like
> 
> rzip <-  function(n, lambda, zip=0.0) {
> 	#zip is the desired proportion of _excess_ zeros
>                if(zip>1.0 || zip <0) stop("zip must be in (0,1)")
>                n.zip <- ceiling(zip*n)
>                n.pois <- n-n.zip
>               sample( c( rpois(n.pois, lambda), rep(0,n.zip) ) )
> }
> 
> #Example
> rzip(50, 3.5, zip=0.5)
> 
> Having said that, I'd bet there's a package out there that already does
> it better...
> 
>  S Ellison
> 
> >
> >
> >
> > Dear Mr/Mrs
> >
> > I am Lili Puspita Rahayu, student from magister third level
> > of Statistics in Bogor Agriculture University.
> > Mr/
> > Mrs, now I'm analyzing the Zero inflated Poisson (ZIP), which
> > is a solution of the Poisson regression where the response
> > variable (Y) has zero excess. ZIP now I was doing did not use
> > real data, but using simulated data in R. Simulations by
> > generating data on variables x1, x2, x3 with each size n =
> > 100, after which generate data on response variable (Y).
> > However, when I generate the variable y, after generating
> > variables x1, x2, x3, then the simulation result in the
> > variable y that does not have a zero excess. Sometimes just a
> > coincidence there are 23%, 25% the proportion of zero on the
> > variable Y. This is because I generate variables x1, x2, x3
> > with a distribution that has a small parameter values​​. I've
> > been consulting with my lecturer, and suggested to generate
> > variable Y that can control the proportion of zero on ​​ZIP
> > analysis. I've been trying to make the syntax, but has not
> > succeeded.I would like to ask for assistance to R to make the
> > syntax to generate simulated Y variables that can control the
> > proportion of zeros after generating variables x1, x2, x3 on
> > ZIP analysis.Thus, I can examine more deeply to determine how
> > much the proportion of zeros on response variable (Y) that
> > can be used in the Poisson regression analysis, parametric
> > ZIP and ZIP semiparametric.
> >
> > syntax that I made previously by generating variable y
> > without being controlled to produce zero excess in R :
> >
> > > b0=1.5
> > > b1=-log(2)
> > > b2=log(3)
> > > b3=log(4)
> > > n=100
> > > x1<-rnorm(n, mean=5, sd=2)
> > > x2<-runif(n, min=1, max=2)
> > > x3<-rnorm(n, mean=10, sd=15)
> > >
> > > y<-seq(1,n)
> > > for(i in 1:n)
> > + {
> > + m<-exp(b0+b1*x1[i]+b2*x2[i]+b3*x3[i])
> > + yp<-rpois(1,m)
> > + y[i]<-yp
> > + }
> >
> >
> > I am very
> > grateful for the assistance of R.
> > I am looking forward to hearing from you. Thank you very much.
> >
> > Sincerely yours
> > Lili Puspita Rahayu
> > 	[[alternative HTML version deleted]]
> >
> >
> 
> *******************************************************************
> This email and any attachments are confidential. Any use, copying or
> disclosure other than by the intended recipient is unauthorised. If
> you have received this message in error, please notify the sender
> immediately via +44(0)20 8943 7000 or notify postmaster at lgcgroup.com
> and delete this message and any copies from your computer and network.
> LGC Limited. Registered in England 2991879.
> Registered office: Queens Road, Teddington, Middlesex, TW11 0LY, UK
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-
> guide.html
> and provide commented, minimal, self-contained, reproducible code.


More information about the R-help mailing list