similarity: Calculate similarity matrices

In CAMCOS/camcos2017: CAMCOS Spring 2017: Efficient and Scalable Spectral Clustering for Document Clustering.

Description

Note: this function sets the self-similarity diagonal = 0. For certain applications, diagonal = 1 may be more appropriate.

Usage

similarity(data, method, rowscaling = NULL, colscaling = NULL,
  sigma = NULL, centers = NULL, seed = NULL, distance = NULL,
  sparse = T)

Arguments

data	Matrix.
method	One of the following: "correlation", "cosine", "dotproduct", "Gaussian".
rowscaling	One of the following: "L2", "L1".
colscaling	Single option: "standardize" (center and divide by variance).
sigma	Variance for Gaussian Kernel method.
centers	Proportion of columns to be sampled, if desired (e.g. centers = .25).
seed	For sampling reproducibility.
distance	For use with Gaussian Kernel; options = "JSdivergence", "L1", "L2"
sparse	Logical. Is the data sparse or not.

Value

Similarity matrix.

See Also

colweights, clustering

CAMCOS/camcos2017 documentation built on May 6, 2019, 9:23 a.m.