outlier {vegsoup} | R Documentation |
Identifies outliers based on mean inter-plot dissimilarity or nearest neighbour criterion. Nearest neighbours can
## S4 method for signature 'Vegsoup' outlier(obj, type = c("mccune", "wildi"), thresh = 0.2, ...)
obj |
A |
type |
See ‘Details’. |
thresh |
Threshold nearest neighbour distance for outliers. If specified and argument
|
... |
Not used. |
For type "mccune"
, the default, average Bray-Curtis dissimilarity of an
outlier plot to other plots is greater than two standard deviations from the mean inter-plot
dissimilarity (McCune & Grace 2002). For type "wildi"
a threshold value
has to be specified (Wildi 2013).
Returns a logical vector equal to the number of plots where outlieres are TRUE
.
If type="wildi"
nearest neighbour distances are given as attribute. See
‘Examples’.
Roland Kaiser using code from function outlier
of package dave.
McCune, B. & Grace, J.B. 2002. Analysis of ecological communities. MjM Software design. Gleneden Beach OR, US.
Wildi, O. 2013. Data Analysis in Vegetation Ecology. 2nd ed. Wiley-Blackwell, Chichester.
plotPCO
, coldiss
, outlier
in dave.
require(vegsoup) data(barmstein) x <- barmstein ( o1 <- outlier(x) ) ( o2 <- outlier(x, type = "wildi") ) # access nearest neighbour distances attr(o2, "distances") # argument thresh triggers type "wildi" o3 <- outlier(x, thresh = 0.3) # drop outliers dim(x) dim(x[!o3, ]) # outliers marked in diagnostic plots plotPCO(x) plotPCO(x, thresh = 0.2)