cluster.assign.torus: Clustering by connected components of ellipsoids

Description

cluster.assign.torus returns clustering assignment for data given icp.torus objects, which can be constructed with icp.torus.

plot.clus.torus plots clustering results, which is given by cluster.obj object, with some options.

Usage

cluster.assign.torus(icp.object, data = NULL, level = NULL)
# S3 method for cluster.obj
plot(
  x,
  assignment = c("outlier", "log.density", "posterior", "mahalanobis"),
  overlay = FALSE,
  out = FALSE,
  ...
)

Arguments

icp.object

an object must be an icp.torus object, which contains all values to compute the conformity score constructed with icp.torus, or a hyperparam.torus object which is generated by hyperparam.torus.

data

n x d matrix of toroidal data on $[0, 2\pi)^d$ or $[-\pi, \pi)^d$. If data = NULL, then data within the icp.object is used.

level

a scalar in $[0,1]$. If argument icp.object is an icp.torus object, the default value for level is 0.1. If argument icp.object is a hyperparam.torus object and level = NULL, then level is set as the optimal level hyperparam.torus$alphahat.

cluster.obj object

assignment

A string. One of "outlier", "log.density", "posterior", "mahalanobis". Default is "outlier".

overlay

A boolean index which determines whether plotting ellipse-intersections on clustering plots. Default is FALSE.

out

An option for returning the ggplot object. Default is FALSE.

...

additional parameter for ggplot2::ggplot()

Value

clustering assignment for data, given icp.torus objects

cluster.id.by.log.density: cluster assignment result based on approximate log-density.
cluster.id.by.posterior: cluster assignment result based on the posterior probability.
cluster.id.outlier: cluster assignment result which regards data not included in conformal prediction set as outliers.
cluster.id.by.Mah.dist: cluster assignment result based on Mahalanobis distance.
level: used level which determines the size of clusters(conformal prediction set).
data: input data which are assigned to each cluster.
icp.torus: icp.torus object which is used for cluster assignment.

References

Jung, S., Park, K., & Kim, B. (2021). Clustering on the torus by conformal prediction. The Annals of Applied Statistics, 15(4), 1583-1603.

Gilitschenski, I., & Hanebeck, U. D. (2012, July). A robust computational test for overlap of two arbitrary-dimensional ellipsoids in fault-detection of kalman filters. In 2012 15th International Conference on Information Fusion (pp. 396-401). IEEE.

Examples

Run this code

# NOT RUN {
data <- toydata1[, 1:2]
icp.torus <- icp.torus(data, model = "kmeans",
                       kmeansfitmethod = "general",
                       J = 4, concentration = 25)
level <- 0.1

cluster.assign.torus(icp.torus, level = level)
# }