Geography Reference
In-Depth Information
Table 13.2
Internal validity
Trip purposes reported by respondents
Trip purposes
predicted by
models
Missing
land-use
data cases
Number
of cases
Accuracy
rates (%)
C1
C2
C3
C4
C5
C6
C7
C1: home
109
9
118
92.4
C2: work
18
1
2
5
26
69.2
C3: school
21
3
24
87.5
C4: shopping
45
3
1
1
7
57
79.0
C5: recreation
1
13
94
108
87.0
C6: personal affairs
4
1
23
19
46
21
114
40.3
C7: Volunteering/
religion
1
2
3
66.7
Table 13.3
External validity
Trip purposes reported by respondents
Trip purposes
predicted by
models
Missing
land-use
data cases
Number
of cases
Accuracy
rates (%)
C1
C2
C3
C4
C5
C6
C7
C1: home
64
5
2
1
72
88.9
C2: work
29
4
2
35
82.9
C3: school
8
8
100
C4: shopping
18
2
3
23
78.3
C5: recreation
1
1
10
59
71
83.1
C6: personal affairs
1
5
9
17
8
40
42.5
C7: Volunteering/
religion
a relative low percentage of personal affairs are correctly detected by the models.
As indicated in the table, a major reason is that there are many cases that the land
use data for personal affairs are missing. From this point of view, this is caused by
availability of data, not by the algorithm. In a way similar to Table 13.2 ,Table 13.3
reports the external validity. Comparing these two tables, one may easily find out
that the external validity is very much comparable to the internal validity. Though
for some cases, the external validity is lower than the internal validity, but in some
other cases, the external validity is even higher than the internal validity.
Overall, the internal and external validities show that the algorithm and data
mining approach proposed in this study can detect activity types and trip purposes
with accuracy rates reasonably good and comparable to that of other studies. Though
for some trip purposes the accuracy rate of detection are not very high, we believe
that it is very much caused by data availability rather than the algorithm.
Search WWH ::




Custom Search