On Wed, Jul 27, 2011 at 2:09 PM, David Winsemius <dwinsem...@comcast.net>wrote:
> > On Jul 27, 2011, at 5:36 AM, Rainer M Krug wrote: > > Sorry for re-iterating - but are there any suggestions on how I could >> tackle >> this problem? >> > > You could start by providing an operational definition for "identity the > areas where d.all is significantly larger then d.co and where it is > significantly smaller" when presumably working from a perspective off not > knowing anything about the parent distributions for the random draws. Let me give you some background on my data and how it was obtained: The data is based on monte carlo simulations of different parameter sets, where the parameter sets are based on a latin hypercube design. The first dataset (dat1) is one input parameter for the simulations, i.e. more or less same probability density for a given range (lets say 1 to 100). Now these simulations predict the population sizes of two species, and the second dataset (dat2) are the same parameter, but for which the smulation predicts co-existence. When deternining the pdfs, I have for some ranges higher densities for dat2 then dat1, i.e. this range favours co-existence, while in others less, i.e. dis-favours co-existence. At the moment, I am just plotting the difference, to indicate in which parameter range co-existence is favoured, but I would like to have a statistical measure to back this up, to avoid the argument "these differences are random effects based on sample size". Eventually, I would be able to draw a ine along the x-axis, which is green if the difference is significantly positive, and negative if significantly negative, and whicte if not different. But is this possible, and how could I do that? > I had no notion of piecewise significance of differences in densities and > figured you were the the one advancing the notion, so you should be defining > what you meant. As I said, I have no idea if something like this can be done - probably moving window type analysis? > I was expecting one of my statistical betters to step in can correct or > comment on the statistical issues, but that hasn't yet happened, so if you > have an operational definition, we could try to implement it across the > range of the two empiric densities. > What I was thinking about, is some kind of moving window analysis, and then using e.g. a ch-square test, to derive a p value to quantify the differences in this window? But what about the window size? Hope this clarifies my question, Rainer > > -- > David > >> >> Thanks, >> >> Rainer >> >> On Tue, Jul 26, 2011 at 2:58 PM, Rainer M Krug <r.m.k...@gmail.com> >> wrote: >> >> Hi >>> >>> this might be a little bit off topic, but here it goes: lets assume I >>> have >>> the following: >>> >>> set.seed(13) >>> dat1 <- rnorm(2000, mean=10, sd=10) >>> dat2 <- rnorm(100, mean=10, sd=20) >>> d.all <- density(dat, n=1024) >>> d.co <- density(x[[v]], , from=min(d.all$x), to=max(d.all$x), >>> n=1024) >>> d.diff <- list( >>> x = d.all$x, >>> y = d.all$y - d.co$y >>> ) >>> >>> ylim <- range(c(d.all$y, d.co$y, d.diff$y)) >>> plot( >>> d.all, >>> ylim = ylim >>> ) >>> abline(h=0) >>> lines(d.co, col="red") >>> lines(d.diff$x, d.diff$y, col="blue") >>> >>> Now I would like to identify the areas where d.all is significantly >>> larger >>> then d.co and where it is significantly smaller. >>> >>> What is the easiest approach to do this? At the moment I am not doing any >>> tests, but I am sure there is a way to determine the ranges >>> statistically? >>> >>> Thanks, >>> >>> Rainer >>> >>> -- >>> Rainer M. Krug, PhD (Conservation Ecology, SUN), MSc (Conservation >>> Biology, >>> UCT), Dipl. Phys. (Germany) >>> >>> Centre of Excellence for Invasion Biology >>> Stellenbosch University >>> South Africa >>> >>> Tel : +33 - (0)9 53 10 27 44 >>> Cell: +33 - (0)6 85 62 59 98 >>> Fax (F): +33 - (0)9 58 10 27 44 >>> >>> Fax (D): +49 - (0)3 21 21 25 22 44 >>> >>> email: rai...@krugs.de >>> >>> Skype: RMkrug >>> >>> >>> >> >> -- >> Rainer M. Krug, PhD (Conservation Ecology, SUN), MSc (Conservation >> Biology, >> UCT), Dipl. Phys. (Germany) >> >> Centre of Excellence for Invasion Biology >> Stellenbosch University >> South Africa >> >> Tel : +33 - (0)9 53 10 27 44 >> Cell: +33 - (0)6 85 62 59 98 >> Fax (F): +33 - (0)9 58 10 27 44 >> >> Fax (D): +49 - (0)3 21 21 25 22 44 >> >> email: rai...@krugs.de >> >> Skype: RMkrug >> >> [[alternative HTML version deleted]] >> >> ______________________________**________________ >> R-help@r-project.org mailing list >> https://stat.ethz.ch/mailman/**listinfo/r-help<https://stat.ethz.ch/mailman/listinfo/r-help> >> PLEASE do read the posting guide http://www.R-project.org/** >> posting-guide.html <http://www.R-project.org/posting-guide.html> >> and provide commented, minimal, self-contained, reproducible code. >> > > David Winsemius, MD > West Hartford, CT > > -- Rainer M. Krug, PhD (Conservation Ecology, SUN), MSc (Conservation Biology, UCT), Dipl. Phys. (Germany) Centre of Excellence for Invasion Biology Stellenbosch University South Africa Tel : +33 - (0)9 53 10 27 44 Cell: +33 - (0)6 85 62 59 98 Fax (F): +33 - (0)9 58 10 27 44 Fax (D): +49 - (0)3 21 21 25 22 44 email: rai...@krugs.de Skype: RMkrug [[alternative HTML version deleted]] ______________________________________________ R-help@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.
