Mining Matrix Pattern from Mobile Users - Distributed Artificial Intelligence, Agent Technology, and Collaborative Applications - page 253

Figure 19. Systematic setup performance evaluation

Size

250

6

5

200

4

150

3

100

2

50

1

0

0

15 20 25

15 20 25

Array Size

Y Axis Size

Percentage of Occupance

8%

7%

6%

5%

4%

3%

2%

1%

0%

15 20 25

Array Size

Figure 18 shows the comparisons among the single layer and multi-layer matrix patterns, each graph
represents different datasets, namely, Equal, Double and Triple respectively. It can be observed that
the relative differences among the net location identifiers were great, and the greater the resolution, the
better the capability for fine tuning the system. However, if the net location identifiers were too small,
it would mean that each location identifier is for a relatively large amount of geographical space, and it
becomes harder to differentiate among the individual physical locations.
Figure 19 shows charts produced from performance evaluation through the systematic setup of test
drawn from data of real life in order to generate a non random scenario. Figure 19 shows the chart rep-
resenting the count of significant elements in a single layer matrix of size 15 * 15, 20 * 20 and 25 * 25.
During this test, the number of significant elements, which are elements having frequency of higher than
the frequency threshold is high at 15 * 15, but gradually reduced to constant at 20 * 20 and 25 * 25.
The second chart shows the total elements in the array of 15 * 15, 20 * 20 and 25 * 25. The purpose
of this chart is to produce a representation of their differences in size, and their rate of increase for each
test case. An appreciation of the size of array is essential for the interpretation of data in performance
evaluation. It tells how much possible empty spaces are there to be filled up to become significant ele-
ment through non random process.
The third chart is a comparison chart for the percentage of occupancy of the significant elements
in the array. A higher percentage of occupancy represents that there are more significant elements
than empty elements. Note that the chart shows different rate of decreasing percentage of occupancy.
Although the general trend is decreasing, but the rate of decrease is higher between X = 15 to X = 20,
than X = 20 to X = 25.