TSClusters-class function

Class definition for TSClusters and derived classes

Formal S4 classes for time-series clusters. See class hierarchy and slot organization at the bottom . class

Details

The base class is TSClusters. The 3 classes that inherit from it are: PartitionalTSClusters, HierarchicalTSClusters and FuzzyTSClusters.

HierarchicalTSClusters also contain stats::hclust() as parent class.

Package clue is supported, but generics from flexclust are not. See also TSClusters-methods .

Slots

• call: The function call.
• family: An object of class tsclustFamily .
• control: An appropriate control object for tsclust(). See tsclust-controls .
• datalist: The provided data in the form of a list, where each element is a time series.
• type: A string indicating one of the supported clustering types of tsclust().
• distance: A string indicating the distance used.
• centroid: A string indicating the centroid used.
• preproc: A string indicating the preprocessing used.
• k: Integer indicating the number of desired clusters.
• cluster: Integer vector indicating which cluster a series belongs to (crisp partition). For fuzzy clustering, this is based on distance , not on fcluster. For hierarchical, this is obtained by calling stats::cutree() with the given value of k.
• centroids: A list with the centroid time series.
• distmat: If computed, the cross-distance matrix.
• proctime: Time during function execution, as measured with base::proc.time().
• dots: The contents of the original call's ellipsis (...).
• args: The contents of the original call's args parameter. See tsclust_args().
• seed: The random seed that was used.
• iter: The number of iterations used.
• converged: A logical indicating whether the function converged.
• clusinfo: A data frame with two columns: size indicates the number of series each cluster has, and av_dist indicates, for each cluster, the average distance between series and their respective centroids (crisp partition).
• cldist: A column vector with the distance between each series in the data and its corresponding centroid (crisp partition).
• method: A string indicating which hierarchical method was used.
• fcluster: Numeric matrix that contains membership of fuzzy clusters. It has one row for each series and one column for each cluster. The rows must sum to 1. Only relevant for fuzzy clustering.

TSClusters

The base class contains the following slots:

• call
• family
• control
• datalist
• type
• distance
• centroid
• preproc
• k
• cluster
• centroids
• distmat
• proctime
• dots
• args
• seed

PartitionalTSClusters

This class adds the following slots to the base class:

• iter
• converged
• clusinfo
• cldist

HierarchicalTSClusters

This class adds the following slots to the base class:

• method
• clusinfo
• cldist

FuzzyTSClusters

This class adds the following slots to the base class:

• iter
• converged
• fcluster

See Also

TSClusters-methods