Skip to main content

Table 1 Taxonomy of visualization techniques for visualizing clustering results.

From: XCluSim: a visual analytics tool for interactively comparing multiple clustering results of bioinformatics data

Principle to Show Cluster Membership

Visualization Component

Clustering Algorithm

Main

Secondary

 

Hierarchical

Partitional

Density-based

Similarity

(color or size)

Proximity

Scatterplot + Color

Δ

Δ [17]

Δ [16]

  

*Graph (vertex as item) + Color

Δ

Δ [5]

Δ [16]

  

*Bar chart (Reachability Plot)

X

X

O [15, 20, 24]

 

Enclosure

Colored shape

Δ [25]

Δ

Δ [13]

 

.

*Parallel coordinate plot + Color

Δ [6]

Δ

Δ

Proximity

Similarity and Enclosure

Bar chart (Silhouettes Plot)

Δ

O [5, 14]

O

Connectedness

(line connection)

Similarity and Proximity

*Dendrogram

O [2, 8, 9]

X

Δ

  

Normal tree

Δ [26]

X

Δ

  

Circular tree

Δ [26]

X

Δ

Enclosure

Proximity

*Heatmap + Partitioning

O

O

O

 

Similarity and Proximity

Treemap

Δ [26]

X

X

*Visualization techniques supported in XCluSim

  1. Visualization components for visualizing clustering results use visual cues based on Gestalt principles of grouping [27] to represent cluster membership. We categorize the visualization components by principle and indicate how appropriate each visualization component is for showing clustering results by different types of clustering algorithms. (i.e. "O" for most appropriate, "Δ" for moderately appropriate, and "X" for not applicable).