ProbKMA: ProbKMA Class

In funMoDisco: Motif Discovery in Functional Data

ProbKMA Class

Description

The 'ProbKMA' class is an R wrapper for the C++ implementation of the Probabilistic K-means Algorithm (ProbKMA) with local alignment. This class facilitates local clustering of functional data and functional motif discovery, as proposed in the paper 'Probabilistic K-means with local alignment for clustering and motif discovery in functional data', authored by Marzia A. Cremona and Francesca Chiaromonte.

Value

A 'ProbKMA' object from the C++ ProbKMA class.

Constructor

Create a 'ProbKMA' object using the following constructor:

prok <- new(ProbKMA, data$Y, data$V, params, data$P0, data$S0, "H1")

Parameters

Y

A list containing functional data and possibly derivatives.

params

An instance of the Parameters class, containing algorithm settings.

P0

A matrix representing the initial membership probabilities.

S0

A matrix representing the initial shift warping parameters.

diss

A character string specifying the dissimilarity measure. Possible choices are:

• ''d0_L2''

• ''d1_L2''

• ''d0_d1_L2''

Usage

You can access and modify the 'ProbKMA' object with the following methods:

Getters:

prok$get_parameters()

Returns a list of parameters.

prok$get_motifs()

Returns a list containing the motifs found.

Setters:

prok$set_P0(P)

Sets the membership matrix.

prok$set_S0(S)

Sets the shift warping matrix.

prok$set_parameters(param)

Sets parameters field by passing a list of parameters.

Initialize Motifs:

prok$reinit_motifs(c, d)

Reinitializes (empty) K motifs with dimension c_k x d.

Run ProbKMA algorithm:

prok$probKMA_run()

Runs the algorithm.

Author(s)

Niccolò Feresini and Riccardo Lazzarini

funMoDisco documentation built on April 16, 2025, 1:10 a.m.