BN_clustering: Clustering Gaussian Bayesian network

In ComputationalIntelligenceGroup/3DSomaMS: 3DSomaMS: A univocal definition of the neuronal soma morphology using Gaussian mixture models

Description

Given a Bayesian network structure, apply clustering to the data with the aim to find clusters of somas

Usage

BN_clustering(bayesian_structure, data, clusters, initialization, verbose)

Arguments

bayesian_structure	a bn object from bnlearn package whose nodes are gaussians
data	the same dataset used to compute the bayesian_structure
clusters	array of values where each value indicates the number of clusters for that model
initialization	a natural number to initialize the algorithm. Use 0 to use kmeans initialization and any other positive number as a seed for a random initialization
verbose	show BIC scores for each model and the optimal number of clusters

Value

model_list is the list of parameters that best fit the data according to BIC criteria

Examples

somas <- read.table(file.path(tempdir(),"somaReebParameters.csv"), sep = ";", header = T, row.names = 1)
bayesian_structure <- BN_structure_learning(path_to_csv = file.path(tempdir(),"somaReebParameters.csv"), nboots = 200, significant_arcs = 0.95)
fittest_model <- BN_clustering(bayesian_structure, data = somas, clusters = c(2, 3, 4), initialization = 2, T)

ComputationalIntelligenceGroup/3DSomaMS documentation built on May 6, 2019, 12:49 p.m.