Setup SpectralLocalConfig

Setup a SpectralLocalConfig object for spectral clustering with a locally scaled Gaussian kernel, via the 'kernlab' package. Each case's kernel width is taken from the distance to its seventh nearest neighbor, which lets one clustering hold groups of differing density. The per-case widths need every pairwise distance, so this variant has no Nystrom approximation.

Usage

setup_SpectralLocal(k = 3L, iterations = 200L, features = NULL)

Arguments

k

Integer [2, Inf): Number of clusters. Must be no more than the number of cases.

iterations

Integer [1, Inf): Maximum number of k-means iterations run on the spectral embedding.

features

Optional Character vector: Names of at least 2 distinct feature columns to cluster on. NULL clusters on all numeric features.

Value

SpectralLocalConfig object.

Details

Argument names are rtemis' own. They map to kernlab::specc() as: k -> centers, with kpar = "local"; iterations keeps its name.

References

Zelnik-Manor L, Perona P (2004). Self-Tuning Spectral Clustering. Advances in Neural Information Processing Systems, 17, 1601-1608.

Author

EDG

Examples

setup_SpectralLocal(k = 3L)

#> <SpectralLocal ClusteringConfig>
#>          k: <int> 3
#> iterations: <int> 200