Segmentation of Continuous Data Streams Based on a Change Detection Methodology - Advanced Techniques in Knowledge Discovery and Data Mining - page 118

105%

100.00% 100.00% 100.00% 100.00%

100%

100.00%

98.63%

95%

92.42% 90%

85%

84.47%

80%

75%

2 3 4 5

Segment

CD XP
Fig. 4.3. Implementation of CD and XP on trial number 1 in “Stock” Database.

120%

100.00% 100.00% 100.00%

100%

99.67% 99.33%

80%

81.20%

60%

40%

20%

0%

2 3 4

Segment

CD XP

Fig. 4.4. Implementation of CD and XP on trial number 2 in “Stock” Database.

The second trial is a partition of 5000 accumulated records into four equally
sized data sets. The assumption for this kind of segmentation, as opposed to the
first one, is that the more data the segment holds, the better the incremental data-
mining model is. Figure 4.4 describes the outcome of implementing the change-
detection methodology on these segments of data.
Opposed to the first two trials, the third trial is based on the assumption that the
first segment should hold enough information to validate the base data-mining