Hi JAmes, I'm afraid I've never used the packages so I'm no expert. But in answer to your questions.. 1. Kulldorf recommends a fract pop of 0.5 but I guess if you have a reason to believe there is a different upper limit then you could compare results. Remember, this is the fraction of the total population, not the population of cases 2. Never used the opgam command..3. I think the kulldorf command in SpatialEpi package can detect more than 1 cluster.. Good luck! > From: roone...@tcd.ie > To: r-sig-geo@r-project.org > Date: Mon, 26 Aug 2013 22:23:26 +0100 > Subject: Re: [R-sig-Geo] output data from R to SaTScan > > Hi Hugh, > > Thanks for your answer. Yes I've seen it said that those packages can do most > of what SaTScan does. > I'm playing with Kulldorff's Statistic as implemented in section 11.5.6 of > ASDAR - but a few things confuse me: > > 1. How do I decide what the correct value for fractpop is ? I initially had > it set to .25 and I was getting cluster of 50% of my cases which made no > sense. > 2. Is there any correction for multiple testing in the opgam() command ? I > have over 3000 areas - do I need to set a very low alpha ? > 3. How do I detect more than one cluster ? I have maps where I suspect 2 or 3 > separate clusters based on smoothed RR's. > > > My code currently looks like this (which I based on the code for Fig11.18 in > the book and this useful question and answer: > http://r-sig-geo.2731867.n2.nabble.com/Strange-results-in-DCluster-package-td7326832.html > ) > > sa<-data.frame(Observed=ED$var1) > sa<-cbind(sa, Expected=mean(ED$var1)) > sa<-cbind(sa, x=ED$x, y=ED$y) > sa$Observed<-as.numeric(sa$Observed) > > #Kuldorff-Nagarwalla analysis > mle<-calculate.mle(sa, model="poisson") > knres<-opgam(data=sa, thegrid=sa[,c("x", "y")], alpha=.025, R=99, > iscluster=kn.iscluster, fractpop=.10, model="poisson", mle=mle, > log.v=TRUE) > > #Print cluster result > clusters<-get.knclusters(as(ED, "data.frame"), knres) > i<-which.max(knres$statistic) > > ED$KNcluster<-"" > ED$KNcluster[clusters[[i]]]<-"cluster" > ED$KNcluster[clusters[[i]][1]]<-"centre" > ED$KNcluster<-as.factor(ED$KNcluster) > > print(spplot(ED, "KNcluster", main="Kulldorff's method", > col.regions=c(gray(1), gray(.5), gray(.8)))) > > > Thanks, > James > > > ________________________________________ > From: Hugh Sturrock [hughsturr...@hotmail.com] > Sent: 26 August 2013 17:10 > To: James Rooney; r-sig-geo@r-project.org > Subject: RE: [R-sig-Geo] output data from R to SaTScan > > Hi James, > > Not sure how to do that, but check out SpatialEpi and Dcluster packages, I > believe they can do a lot of what SaTScan does.. > > Cheers, Hugh > > > From: roone...@tcd.ie > > To: r-sig-geo@r-project.org > > Date: Mon, 26 Aug 2013 16:42:55 +0100 > > Subject: [R-sig-Geo] output data from R to SaTScan > > > > Hi all, > > > > So I have created some maps in R, smoothed then using the BYM model in > > Openbugs and opened the CODA files in R once again to analyze results etc. > > > > I would like to now test for clusters using SaTScan and the post smoothing > > data - but I can find very little info anywhere on how to output my R data > > into a format that SaTScan can use. My data within R is in the form of a > > SpatialPolygonsDataFrame which up until now I have been saving as an ESRI > > shapefile with the command: > > writeOGR(data, "", layer="data_out",driver="ESRI Shapefile",overwrite=T) > > > > But from what info I can find, I think this doesn't seem to be a useful > > format for SaTScan. > > > > Any advice would be appreciated on how to achieve this! > > Many thanks, > > James > > _______________________________________________ > > R-sig-Geo mailing list > > R-sig-Geo@r-project.org > > https://stat.ethz.ch/mailman/listinfo/r-sig-geo > > _______________________________________________ > R-sig-Geo mailing list > R-sig-Geo@r-project.org > https://stat.ethz.ch/mailman/listinfo/r-sig-geo [[alternative HTML version deleted]]
_______________________________________________ R-sig-Geo mailing list R-sig-Geo@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-geo
