clustHVT: Performing Hierarchical Clustering Analysis

In HVT: Constructing Hierarchical Voronoi Tessellations and Overlay Heatmaps for Data Analysis

Performing Hierarchical Clustering Analysis

Description

This is the main function to perform hierarchical clustering analysis which determines optimal number of clusters, perform AGNES clustering and plot the 2D cluster hvt plot.

Usage

clustHVT(
  data,
  trainHVT_results,
  scoreHVT_results,
  clustering_method = "ward.D2",
  indices,
  clusters_k = "champion",
  type = "default",
  domains.column
)

Arguments

data	Data frame. A data frame intended for performing hierarchical clustering analysis.
trainHVT_results	List. A list object which is obtained as a result of trainHVT function.
scoreHVT_results	List. A list object which is obtained as a result of scoreHVT function.
clustering_method	Character. The method used for clustering in both NbClust and hclust function. Defaults to ‘ward.D2’.
indices	Character. The indices used for determining the optimal number of clusters in NbClust function. By default it uses 20 different indices.
clusters_k	Character. A parameter that specifies the number of clusters for the provided data. The options include “champion,” “challenger,” or any integer between 1 and 20. Selecting “champion” will use the highest number of clusters recommended by the ‘NbClust’ function, while “challenger” will use the second-highest recommendation. If a numerical value from 1 to 20 is provided, that exact number will be used as the number of clusters.
type	Character. The type of output required. Default is 'default'. Other option is 'plot' which will return only the clustered heatmap.
domains.column	Character. A vector of cluster names for the clustered heatmap. Used only when type is 'plot'.

Value

A list object that contains the hierarchical clustering results.

[[1]]	Summary of k suggested by all indices with plots
[[2]]	A dendogram plot with the selected number of clusters
[[3]]	A 2D Cluster HVT Plotly visualization that colors cells according to clusters derived from AGNES clustering results. It is interactive, allowing users to view cell contents by hovering over them

Author(s)

Vishwavani <[e-mail]>

Examples

data("EuStockMarkets")
dataset <- data.frame(t = as.numeric(time(EuStockMarkets)),
                     DAX = EuStockMarkets[, "DAX"],
                     SMI = EuStockMarkets[, "SMI"],
                     CAC = EuStockMarkets[, "CAC"],
                     FTSE = EuStockMarkets[, "FTSE"])
rownames(EuStockMarkets) <- dataset$t
hvt.results<- trainHVT(dataset[-1],n_cells = 30, depth = 1, quant.err = 0.1,
                      distance_metric = "L1_Norm", error_metric = "max",
                      normalize = TRUE,quant_method = "kmeans")
scoring <- scoreHVT(dataset, hvt.results, analysis.plots = TRUE, names.column = dataset[,1])
centroid_data <- scoring$centroidData
hclust_data_1 <- centroid_data[,2:3]
clust.results <- clustHVT(data = hclust_data_1, 
                         trainHVT_results = hvt.results,
                         scoreHVT_results = scoring, 
                         clusters_k = 'champion', indices = 'hartigan')

HVT documentation built on April 3, 2025, 8:45 p.m.