We have applied a nonmetric multidimensional scaling (NMDS) to represent objects (three habitats) in an ordination plot and to find the best representation of objects, NMDS applies an iterative procedure that tries to position the objects in the requested number of dimensions in such a way as to minimize a stress function (scaled from 0 to 1) which measure the goodness-of-fit of the distance adjustment in the reduced-space configuration. Consequently, the lower the stress value, the better the representation of objects in the ordination space is.
Wha't exactly the question, the lower the stress the better the representation.
I always try to stay between a stress vale of 0.1 and 0.2. Increasing the number of random starts in search of a solution normally ensures you get a low stress value. If it doesn't work, then you probably have to consider increasing the number of dimensions. Hope this helps.
The stress will decline with the number of dimensions within which the objects are positioned. I suggest that you examine the decline in stress with increasing number of dimensions. The optimal number of dimensions (hopefully 2 or 3) will occur when the addition of another dimension offers little further reduction. Having decided on the number of dimensions you need to do a few runs with random starting positions to ensure the method is finding the minimal stress. There is no actual stress value which indicates a good ordination. The lower the stress the closer the plotted relationships between objects in the reduced space reflects there actual similarity over all the available dimensions.
Dear Professor Peter,
Many thanks for your clear and detailed answer.
All Best,
Hi Maarten,
Yes, I would like to know the lowest stress value for better ordination in NMDS.
Sincerely,
- Badges
- Science method
- Similar topics
- Biological Science
- Ecology