clusterlab: clusterlab
In clusterlab: Flexible Gaussian Cluster Simulator

Description Usage Arguments Value Examples

View source: R/clusterlab.R

This function runs clusterlab which is a simulator for Gaussian clusters. The default method positions cluster centers on the perimeter of a circle, before creating gaussian clusters around them and projecting the 2D co-ordinates into high dimensional feature space. This method allows control over the spacing, variance, and size of the clusters. Also included is a simple random cluster simulator where the spacing of the clusters cannot be controlled precisely, but the other parameters can.

clusterlab(centers = 1, r = 8, sdvec = NULL, alphas = NULL,
  centralcluster = FALSE, numbervec = NULL, features = 500,
  seed = NULL, rings = NULL, ringalphas = NULL, ringthetas = NULL,
  outliers = NULL, outlierdist = NULL, mode = c("circle", "random"),
  pcafontsize = 18, showplots = TRUE)

`centers`	Numerical value: the number of clusters to simulate (N)
`r`	Numerical value: the number of units of the radius of the circle on which the clusters are generated
`sdvec`	Numerical vector: standard deviation of each cluster, N values are required
`alphas`	Numerical vector: how many units to push each cluster away from the initial placement, N values are required
`centralcluster`	Logical flag: whether to place a cluster in the middle of the rest
`numbervec`	Numerical vector: the number of samples in each cluster, N values are required
`features`	Numerical value: the number of features for the data
`seed`	Numerical value: fixes the seed if you want to repeat results, set the seed to 123 for example here
`rings`	Numerical value: the number of concentric rings to generate (previous settings apply to all ring clusters)
`ringalphas`	Numerical vector: a vector of numbers to push each ring out by, must equal number of rings
`ringthetas`	Numerical vector: a vector of angles to rotate each ring by, must equal number of rings
`outliers`	Numerical value: the number of outliers to create
`outlierdist`	Numerical value: a distance value to move the outliers by
`mode`	Character string: whether to use the standard method (circle), or simple random placement (random)
`pcafontsize`	Numerical value: the font size of the pca
`showplots`	Logical flag: whether to remove the plots

A list, containing: 1) the synthetic data 2) cluster membership matrix

1
2
3

synthetic <- clusterlab(centers=4,r=8,sdvec=c(2.5,2.5,2.5,2.5),   
alphas=c(1,1,1,1),centralcluster=FALSE,   
numbervec=c(50,50,50,50)) # for a six cluster solution)