Calculate CDbw index

Computes the CDbw-index (Halkidi and Vazirgiannis 2008; Halkidi, Vazirgiannis and Hennig, 2015). This function is directly copied from the fpc CRAN package and was written by Christian Hennig. It is included here to reduce the package dependencies (since fpc has a few not-so-light dependencies that aren't required here).

Usage

CDbw(
  x,
  labels,
  r = 10,
  s = seq(0.1, 0.8, by = 0.1),
  clusterstdev = TRUE,
  trace = FALSE
)

Arguments

x

Something that can be coerced into a numerical matrix, with elements as rows.

labels

A vector of integers with length =nrow(x) indicating the cluster for each observation.

r

Number of cluster border representatives.

s

Vector of shrinking factors.

clusterstdev

Logical. If TRUE, the neighborhood radius for intra-cluster density is the within-cluster estimated squared distance from the mean of the cluster; otherwise it is the average of these over all clusters.

trace

Logical; whether to print processing info.

Value

A vector with the following values (see refs for details):

cdbw

value of CDbw index (the higher the better).

cohesion

cohesion.

compactness

compactness.

sep

separation.

References

Halkidi, M. and Vazirgiannis, M. (2008) A density-based cluster validity approach using multi-representatives. Pattern Recognition Letters 29, 773-786.

Halkidi, M., Vazirgiannis, M. and Hennig, C. (2015) Method-independent indices for cluster validation. In C. Hennig, M. Meila, F. Murtagh, R. Rocci (eds.) Handbook of Cluster Analysis, CRC Press/Taylor & Francis, Boca Raton.

Author

Christian Hennig

Examples

d1 <- mockData()
CDbw(d1[,1:2], d1[,3])
#>        cdbw    cohesion compactness         sep 
#>  0.03805049  0.23717624  0.24202068  0.66288255