[R-sig-eco] some MDS and R problems
Dave Roberts
dvrbts at ecology.msu.montana.edu
Thu Jun 17 02:01:52 CEST 2010
Arnaldo,
Given that NMDS operates essentially on rank distances, small
variations in dissimilarity generally have low impact. I rarely have
points with zero dissimilarity on quantitative indices, but it sometimes
happens on presence/absence based indices. If you have a dissimilarity
object named dis, I just do
> dis[dis==0] <- 0.0001
> res <- nmds(dis)
Following Jari's argument about the other point's contribution to
stress, I think fudging one dissimilarity is a better solution than
dropping out a point
Dave Roberts
Arnaldo Russo wrote:
> Hi all,
>
> I'm executing some multivariate exploratory annalysis, with vegan and MASS
> packages.
>
> Some problems are not solved for me at this moment. I have read all past
> discussions over MDS problems.
>
> In my studies occurs a situation that some samples are equal each other, and
> this results in zero dissimilarities.
>
> I tryed with no success, use the stepacross function. So, in this case I
> changed one of these zero dissimilarities values for a minimum (1e-5). Some
> of these modifications are different than "stepacross" or same isoMDS(x.dist
> + 1e-5) (proposed by Jari oksanen to 'lie' to isoMDS? When I delete a
> sample I do not loose some weight?
>
> Changing one of those zero dissimilarities and using zerodist = "add" option
> of metaMDS, it passed the current error (zero or negative distance between
> objects).
>
> Someone could explain the function posted by Gavin Simpson (Jan 12, 2010;
> 05:45pm Re: Non-metric multidimensional scaling (NMDS) help). I didn't get
> some result.
>
> If there is a good reason, and you want to include all samples, then
> you'll need to come up with a means for handling them. metaMDSdist allow
> you to add a small value to the zero dissimilarities. The details are in
>> the code, but effectively all zero distances are replaced by half the
>> smallest non zero distance. You could do a similar replacement yourself
>> if you feel this is warranted and/or justified.
>>
>> minDij <- min(Dij[Dij > 0) / 2
>> Dij[Dij <= 0] <- minDij
>>
>> Will do this replacement if Dij is your matrix (replace Dij with
>> whatever the name of your matrix is). Then supply the new matrix to
>> metaMDS. "
>>
>
>
> When I use unique function as #x.dist <- dist(unique(X))
> this cutted some of my samples that I do not know. I can't see my variables
> in the plot. Some help?
>
> Cheers, Arnaldo.
>
>
> epi<- read.table("epi.txt", h=TRUE,row.names=1)
>
> library('vegan')
> library('MASS')
>
> epi.<- as.matrix(epi)
>
> dissim1<- vegdist(epi., method="jaccard", binary=TRUE)
> dissim1.mds<-metaMDS(dissim1,k=2)
>
> plot(dissim1.mds,type="n")
> text(dissim1.mds, labels=as.character(row.names(epi)))
>
> #epi
> sample sp1 sp2 sp3 sp4 sp5 sp6 sp7 sp8 sp9
> sp10 sp11 sp12 sp13 sp14 sp15 sp16 sp17 sp18 sp19
> 1E 1 0 1 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0
> 2E 0 0 1 0 1 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0
> 3E 1 0 1 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0
> 4E 1 0 1 0 1 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0
> 5E 1 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0
> 6E 1 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0
> 7M 1 0 1 0 0 0 1 0 0 0 0 0 0 0
> 0 0 0 0 0
> 8E 1 0 0 0 0 0 1 0 0 0 0 0 0 0
> 0 0 0 0 0
> 11E 1 0 0 0 1 0 1 0 0 0 0 0 0 0
> 0 0 0 0 0
> 12E 0 0 0 0 0 1 0 0 0 0 0 0 0 0
> 0 0 0 0 0
> 13E 0 0 0 0 0 0 1 0 0 0 1 1 0 0
> 0 0 0 0 0
> 14E 0 0 1 0 0 0 0 0 0 0 0 1 0 0
> 0 0 0 0 0
> 15E 1 0 0 0 0 0 1 0 0 0 0 0 1 0
> 0 0 0 0 0
> 17E 0 0 0 0 1 0 1 0 1 0 0 0 0 0
> 0 0 0 0 0
> 18E 0 0 0 0 0 0 1 0 0 0 0 0 0 0
> 0 0 0 0 0
> 20E 1 0 0 0 1 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0
> 21E 1 0 0 0 0 0 0 0 0 0 0 0 0 1
> 0 0 0 0 0
> 22E 1 0 0 0 1 0 0 0 1 0 0 0 0 0
> 0 0 0 0 0
> 23E 1 0 1 1 1 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0
> 27E 1 0 0 1 1 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0
> 28E 1 0 0 0 1 0 1 1 0 0 1 0 0 0
> 0 0 0 0 0
> 30E 1 1 0 0 1 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0
> 31E 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 1 0 0 0 0
> 32E 1 0 0 0 0 0 0 0 1 0 0 0 0 0
> 0 0 0 0 0
> 9E 0 0 0 0 0 1 1 1 1 1 0 0 0 0
> 0 0 1 0 0
> 10E 0 0 1 0 0 1 1 0 1 1 1 0 0 0
> 0 1 0 1 1
>
> ------
> Arnaldo D`Amaral Pereira Granja Russo
> Instituto Ambiental Boto Flipper
> www.institutobotoflipper.com.br
>
> [[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-ecology mailing list
> R-sig-ecology at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
More information about the R-sig-ecology
mailing list