two-tone pseudo coloring for multiple...

1
Two-Tone Pseudo Coloring for Multiple Variables Overview Goal: Visualization of multivariate data Approach: Combine advantages of Two-Tone Pseudo Coloring (TTPC) TableLens Self-Organizing Maps (SOM) Challenges: TTPC supports only 1-D data TTPC's efficiency decreases for high-frequency data Improve readability of TTPC Visual Layout and Mapping Tabular layout: Relies on the well-known TableLens approach Larger number of data rows can be represented Color Coding: Makes use of TTPC to color-code each table cell Allows for more variables to be visualized within the TableLens SOM Sorting Generate SOM with number of neurons and training phases estimated based on number of variables and data rows Apply SOM to sort the data with respect to all or user-selected variables High frequencies in visual representation are reduced Clusters may be identified Side effect: hierarchical data structure can be used to drive interface for information drill-down TableLens, Rao & Card, 1994 Two-Tone Pseudo Coloring, Saito et al., 2005 Improve Readability Color scale has major impact on efficiency of TTPC Key to comprehensibility: segmentation of color scale Simple heuristics help to generate easier interpretable color scales Mathias John, Christian Tominski, & Heidrun Schumann University of Rostock, Germany Basic color segment computation Heuristics applied Data with high frequencies SOM Sorting

Upload: others

Post on 19-Oct-2020

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Two-Tone Pseudo Coloring for Multiple Variablesvcg.informatik.uni-rostock.de/~ct/pub_files/John06TwoToneTableLe… · variables High frequencies in visual representation are reduced

Two-Tone Pseudo Coloringfor Multiple Variables

OverviewGoal: Visualization of multivariate data

Approach: Combine advantages of● Two-Tone Pseudo Coloring (TTPC)● TableLens● Self-Organizing Maps (SOM)

Challenges:● TTPC supports only 1-D data● TTPC's efficiency decreases for high-frequency data● Improve readability of TTPC

Visual Layout and MappingTabular layout:

● Relies on the well-known TableLens approach● Larger number of data rows can be represented

Color Coding:● Makes use of TTPC to color-code each table cell● Allows for more variables to be visualized within the TableLensSOM Sorting

● Generate SOM with number of neurons and training phases estimated based on number of variables and data rows● Apply SOM to sort the data with respect to all or user-selected variables

● High frequencies in visual representation are reduced● Clusters may be identified

● Side effect: hierarchical data structure can be used to drive interface for information drill-down

TableLens, Rao & Card, 1994 Two-Tone Pseudo Coloring, Saito et al., 2005

Improve Readability● Color scale has major impact on efficiency of TTPC● Key to comprehensibility: segmentation of color scale● Simple heuristics help to generate easier interpretable color scales

Mathias John, Christian Tominski, & Heidrun SchumannUniversity of Rostock, Germany

Basic color segment computation Heuristics applied

Data with high frequencies

SOM Sorting