VORONOI DIAGRAM BASED DIMENSIONAL ANCHOR ASSESSMENT FOR RADIAL VISUALIZATIONS - Information Visualisation: Techniques, Usability and Evaluation - page 267

for a convex polygon is O(log n ) [11] so we have a total of O( n log n ).
Finally, the actual calculation of Q is done in constant time.

Fig. 13.14. Voronoi diagram of the Cars data set with dimensions selected with
mean ratio and arranged in a class discrimination layout (CDL) with AAM
placement. Q score is -0.577 [9].

Dimensions
Selected

Data set Dimensions Classes Q score Remarks

maximum 5
per class

Cars [15] 7 3 -0.577

No classes.
All
dimensions
used.

Htong [15] 4 4 N/A 0.145

DAs

placed
uniformly.

Khan
SRBCT [6]

maximum 5
per class

2308 4 0.619

maximum 7
per class

Spam [16] 57 2 0.380

Transfusion
[17]

maximum 3
per class

5 2 0.789

Table 1. The Q scores of several publicly available data sets [9]. All except
Htong, as noted, used CDL+AAM positioning.