Hi everyone, Have sample of items for each one, a set of 20 dichotomous (absent-present) variables are expressed. I'm trying to understand how to explore the co-occurence of each variable. Read some papers concerning smallest space analysis, but it does not seems implemented in any R package (and my protamming skills are =0). Non metric MDS gives error messages, probably because of the dichotomous character of my variables. My aim is basically to obtain a 2D plot with cooccurence expressed as distance between plots (variables). Any advice to such issue? Thanks in advance and sorry for this long post marco -- View this message in context: http://r.789695.n4.nabble.com/Dichotomous-variables-tp3721550p3721550.html Sent from the R help mailing list archive at Nabble.com.
Dichotomous variables
3 messages · Marco, Sarah Goslee
Hi Marco,
On Fri, Aug 5, 2011 at 11:53 AM, marco <marco.milella at aim.uzh.ch> wrote:
Hi everyone, Have sample of items for each one, a set of 20 dichotomous (absent-present) variables are expressed. I'm trying to understand how to explore the co-occurence of each variable. Read some papers concerning smallest space analysis, but it does not seems implemented in any R package (and my protamming skills are =0). Non metric MDS gives error messages, probably because of the dichotomous character of my variables. My aim is basically to obtain a 2D plot with cooccurence expressed as distance between plots (variables). Any advice to such issue?
I would think that NMDS with a dissimilarity metric appropriate for presence-absence data is a reasonable place to start. Since you don't give any details about your data, or the commands you used to run NMDS, or even the error messages you received, we can't help further. You did calculate a dissimilarity before trying to run NMDS, right? Having binary variables doesn't make a bit of difference. Sarah
Sarah Goslee http://www.functionaldiversity.org
1 day later
Hi Sarah, sorry for the lack of details..hope now my dilemma will be clearer: so, I have my dataset, represented by a matrix:100 subjects and 20 variables, expressed as 1(presence) and 0 (absence) I then tried to obtain a dissimilarity m. using the jaccard() command. To perform a NMDS, I used the isoMDS() command from the vegan package, specifying k=2 dimensions. The returned error sentence says that I have "0 or negative" distance between some subjects (and that's true, a lot have an identical expression of each variable). adding zerodist="ignore" in isoMDS does not solve the problem: same message. So: this is the blind alley. thanks for any feedback marco -- View this message in context: http://r.789695.n4.nabble.com/Dichotomous-variables-tp3721550p3724639.html Sent from the R help mailing list archive at Nabble.com.