Information Technology Reference
In-Depth Information
Data set
Method
AB
Small, random
11.5
11.6
Large, random
27.9
17.1
Small, clustered
9.7
8.2
Large, clustered
24.0
13.5
All documents
49.4
60.1
First 1000
21.1
35.4
Last 1000
1.0
5.5
70
60
Method A
Method B
50
40
30
20
10
0
F IGURE 2. Elapsed time(milliseconds) for methods A andBapplied to data
sets 1-7.
Fig. 11.5 A table compared to a graph. The data shows how two methods compare over seven
experiments. The graph is a better choice for this data because the pattern is more obvious
Graph-drawing tools allow bar graphs to be three dimensional, but the addition
of depth is deceptive; if one bar is twice the height of another, the depth exaggerates
the difference.
Graphs are used to illustrate change in one parameter as another is varied. In some
cases more than two parameters can interact in complex ways. If two parameters,
say B and C , depend on another, A , then a good solution is to plot A on the x -axis
and have two y axes, one for each of B and C , as shown in the graph in Fig. 11.3 .
If two parameters, say D and E , jointly determine a third, F , in some complex
way—thus describing a three-dimensional space—the problem is more difficult. Use
of a three-dimensional graph is an option, in which varying both D and E produces
Search WWH ::




Custom Search