Information Technology Reference
In-Depth Information
•
Similarity based on spatio-temporal single point attributes,
S
SP
The spatio-temporal similarity is calculated based on all single point attribute similari-
ty values, i.e., the average of the flow (
f
), speed (
s
), headway (
h
) and occupancy (
o
)
on all lanes and the standard deviation of the flow (
f
d
), speed (
s
d
), headway (
h
d
) and
occupancy (
o
d
) between the lanes. These eight attributes are determined at each point
along the distance considered and the given time interval as shown in Fig. 3. The
attribute values are normalised between the interval [0,1] based on the range of values
of each attribute found in the available flow data. The distance between corresponding
points between the target case and the case from the case base is given by:
1
(5)
8
,
,
where
x
T,l
is the normalised value of attribute
l= k, s, h, o, k
d
, s
d
, h
d
, o
d
in the target
case and
x
C,l
is the normalised value of attribute
l
in the case from the case base.
l
T
(t
n
, d
n
)
Target Case CT
Case Cc
l
C
(t
n
, d
n
)
Time
T
(Distance
L
)
Fig. 3.
Graph showing values
L
T
(t
n
,d
n
)
and
l
C
(t
n
,d
n
)
of attribute
l
either over time
T
or distance
L
for target case
C
T
and a case from the case base
C
C
. The lines connecting corresponding points
in the attribute value curves represent the difference in the attribute values with respect to time
T
or distance
L.
The aggregate distance
D
avg
is obtained by taking the mean of the difference be-
tween individual point values of the attributes of the target case and a case from the
case base in space and time. The similarity
S
SP
is given by
(6)
1
Similarity based on variation of flow over distance or time interval,
S
var
•
This value refers to the difference between the variation of attribute values flow (
f
)
and speed (
s
) over a distance
L
or a time interval
T
as given in expressions (1) and (2).
The difference in net variation between two cases over a time interval constitutes the
temporal difference
D
temp
and the difference in net change between two cases over a
distance constitutes the spatial difference
D
spatial
. The similarity
S
var
is given by
