clust_matrix: Cluster Earth Mover's Distance Matrix

In MrMaximumMax/FBCanalysis: Develop and evaluate time series data models based on fluctuation based clustering

Cluster Earth Mover's Distance Matrix

Description

Cluster Earth Mover's Distance Square Matrix data and record cluster assignments for involved Patient_IDs for a specified clustering technique and number of clusters.

Usage

clust_matrix(matrix, method, nclust, plotclust)

Arguments

matrix	Object of type matrix storing Earth Mover's Distances for patient time series data distribution pairs
method	Clustering method (hierarchical, kmeans, diana, fanny, som, modelbased, sota, pam, clara)
nclust	Number of clusters (if not specified, user will be asked in the terminal)
plotclust	TRUE/FALSE if clustering data should be visualized (TRUE by default)

Details

Hierarchical clustering describes a general agglomerative hierarchical clustering approach in which the optimum value of an objective function is used to choose which pair of clusters should merge at each step (see hclust for more details).

K-means clustering is a vector quantization technique that divides a set of n observations into k groups, with each observation belonging to the cluster with the closest mean or centroid (see kmeans for more details).

The divisive analytic method (DIANA) constructs a hierarchical clustering structure, starting with a single huge cluster containing all n observations. Clusters are further split until each has a single observation. At each step, the cluster with the largest diameter is selected, where the diameter of a cluster is defined as the biggest difference between any two of its observations (see diana for more details).

While partitioning around medoids (PAM) is comparable to k-means, it is considered more robust since it allows for the use of dissimilarities other than euclidean distance. As with k-means, the number of clusters is determined in advance, and an initial set of cluster centers is required to begin the process (see pam for more details).

Clustering large applications (CLARA) is a system that involves sampling to apply PAM to a sequence of sub-datasets. When the number of observations is big, this results in shorter run times. It is substantially faster than other partitioning algorithms such as PAM at handling huge datasets. Internally, this is performed by examining fixed-size sub-datasets, which results in linear rather than quadratic time and storage requirements for n (see clara for more details).

FANNY explains a fuzzy analysis clustering method. Each observation is distributed through-out the numerous groups in a fuzzy clustering (see fanny for more details).

Self-organizing maps (SOM) are a widespread unsupervised learning technique used by com- putational biologists and academics in machine learning. SOM is a neural network-based system that is well-known for its ability to map and display two-dimensional data (see SOMclustering for more details).

Modelbased clusterinng fits the data to a statistical model composed of a finite mixture of Gaussian distributions. Each mixture component represents a cluster, and the maximum like- lihood method, or estimation maximum (EM), is used to determine the mixture components and group memberships (see ModelBasedClustering for more details).

SOTA, self-organizing tree algorithm, denotes an unsupervised network with a hierarchical and divisive binary tree topology. It is a fast approach, which makes it suitable for clustering a large number of elements. It combines the advantages of hierarchical clustering with those of SOMs. The algorithm chooses the most diverse node and separates it into two nodes referred as cells (see sota for more details).

Value

Object of type list storing cluster data and clustering assignments for the Patient_IDs from the Earth Mover's Distance matrix

References

Joe H Ward Jr. Hierarchical grouping to optimize an objective function. Journal of the American statistical association, 58(301):236–244, 1963.

Stephane Tuffery. Data mining and statistics for decision making. John Wiley & Sons, 2011.

Fionn Murtagh. A survey of recent advances in hierarchical clustering algorithms. The computer journal, 26(4):354–359, 1983.

Fionn Murtagh. Clustering in massive data sets. In Handbook of massive data sets, pages 501–543. Springer, 2002.

Greg Hamerly and Charles Elkan. Alternatives to the k-means algorithm that find better clusterings. In Proceedings of the eleventh international conference on Information and knowledge management, pages 600–607, 2002.

R Wehrens and J Kruisselbrink. kohonen: Supervised and unsupervised self-organising maps r package version 3.0. 10, 2019.

Leonard Kaufman and Peter J Rousseeuw. Finding groups in data: an introduction to cluster analysis. John Wiley & Sons, 2009.

Mark Van der Laan, Katherine Pollard, and Jennifer Bryan. A new partitioning around medoids algorithm. Journal of Statistical Computation and Simulation, 73(8):575–584, 2003.

Chris Fraley and Adrian E Raftery. Model-based clustering, discriminant analysis, and density estimation. Journal of the American statistical Association, 97(458):611–631, 2002.

Javier Herrero, Alfonso Valencia, and Joaquın Dopazo. A hierarchical unsupervised growing neural network for clustering gene expression patterns. Bioinformatics, 17(2):126–136, 2001.

Examples

list <- patient_list(
"https://raw.githubusercontent.com/MrMaximumMax/FBCanalysis/master/demo/phys/data.csv",
GitHub = TRUE)
#Sampling frequency is supposed to be daily
matrix <- emd_matrix(list, "FEV1")
clustering <- clust_matrix(matrix, method = "hierarchical", nclust = 2)

MrMaximumMax/FBCanalysis documentation built on June 23, 2022, 8:21 p.m.