Centroids {NCSampling}

Calculate centroids

Description

Separates a single stratum of the population file into n clusters and finds the centroid of each cluster, where n is the sample size. Not intended to be called directly.

Usage

Centroids(popfile, nrefs, desvars, ctype, imax, nst)

Arguments

popfile

population file - dataframe containing information relating to all plots in the stratum.

nrefs

scalar defining the number of reference plots - required sample size for the stratum.

desvars

character vector containing the names of the design variables.

ctype

clustering type - either k-means ('km') or Ward's D2 ('WD').

imax

maximum number of iterations when calling the k-means clustering procedure.

nst

number of random initial centroid sets when calling the k-means clustering procedure.

Details

The virtual plots are partitioned so as to minimise the sums of squares of distances from plots to cluster centroids. This is done by using a multivariate clustering procedure such as k-means clustering (Hartigan & Wong, 1979) or Ward's D2 clustering (Murtagh & Legendre, 2013), using standardized design variables and a Euclidean distance metric.

Value

centroids

dataframe containing centroids.

cmns

dataframe containing centroid means.

Author(s)

G Melville

References

Hartigan & Wong (1979) Algorithm AS 136: a K-means clustering algorithm. Applied Statistics 28, 100-108, DOI:10.2307/2346830.

Murtagh, M & Legendre, P. (2014) Ward's hierarchical agglomerative clustering method: which algorithms implement Ward's criterion? Journal of Classification, 31, 274-295, DOI: 10.1007/s00357-014-9161-z.

See Also

Existing, NC.sample and kmeans.

Examples

## Centroids(popfile, nrefs, desvars, ctype='km', imax=200, nst=20)