Two-Tone Pseudo-Colored TableLens
A TableLens Featuring Two-Tone Pseudo-Coloring and Hybrid SOM Clustering
Christian Tominski and Mathias JohnUniversity of Rostock, 2006-2008
About
The TableLens by Rao and Card (1994) is an excellent example of a simple, yet expressive visualization method that can help in analyzing even larger volumes of data. This TableLens features two-tone pseudo coloring and a hybrid SOM clustering. Two-tone pseudo coloring can communicate larger volumes of data while still maintaining precision. The hybrid SOM clustering helps to extract and communicate complementary structural information about the data and also serves to drive interactive information drill-down.Publications
- M. John, C. Tominski, and H. Schumann. Visual and Analytical Extensions for the Table Lens. Proceedings of Visualization and Data Analysis (VDA), San Jose, USA, SPIE/IS&T, 2008.
- M. John, C. Tominski, and H. Schumann. Interactive Poster: Two-Tone Pseudo Coloring for Multiple Variables. Poster at IEEE Information Visualization Conference (InfoVis), Baltimore, USA, 2006.
Gallery


Download
You may try out this Java software. Download the file TTTL.zip and if you are lucky you can (still) run it out of the box by calling java -jar TTTL.jar on the command line. On high-resolution displays, you may want to insert -Dsun.java2d.uiScale=2.0 right before the -jar switch with a scaling factor that matches your device configuration (e.g., 2.0 for 200% display scaling). This research prototype is provided "as is" without warranty of any kind.Controls
- Left mouse button: expand/collapse a row, hold and drag to affect multiple rows
- Mouse wheel: scroll
- [SHIFT] + mouse wheel: zoom in/out
- Use toolbar buttons: Open, save, cursor, zoom in, zoom out, color map, color segments, clustering (may take a while)
- Click column header to sort via column
- Right click or [META] + click: Mark column for consideration during SOM clustering
- Drag column to reorder columns
- Click Icicle Plot (appears after clustering) to expand/collapes clusters
- Click Icicle Plot header to expand/collapse whole levels of clustering hierarchy
Data Requirements
- The first line in a data file must contain unique attribute names separated by semicolon
- The second line contains the data types of the attributes also separated by semicolon
- Valid data types are "String", "Real", "Integer", "Category", and "Date"
- The rest of the lines contains the actual data
- Note that a temporal reference is only recognized if "Date" is the first attribute in the file
- Further note that the tool is not very cooperative with respect to duplicate attribute identifiers, missing values, or too many separators, so be careful when preparing your data
- Example:
CAR;MPG;Cylinders;Horsepower;Weight;Acceleration;Year;Origin String;Integer;Category;Integer;Integer;Real;Integer;Category CAR1;18;8;130;3504;12;70;A CAR2;15;8;165;3693;11.5;70;A CAR3;18;8;150;3436;11;70;A CAR4;16;8;150;3433;12;70;B ...