Information Technology Reference
In-Depth Information
n
p
(1)
maxn
a.
Constrained capacity t expresses a relationship between the average item size and the bin
size. The size of item i is s i and the bin size is c .
(s / c)
i
i
t
1

i
n
(2)
n
b.
Item dispersion d expresses the dispersion degree of the item size values.
d
 
()
t
(3)
c.
Number of factors f expresses the proportion of items whose sizes are factors of the bin
capacity.
factor(c ,s )
i
i
f
1

in
(4)
n
d.
Bin usage b expresses the proportion of the total size that can fit in a bin of capacity c .
1if
c
s
i
i
b
1

i
n
(5)
c
otherwise
s
i
i
Prediction phase
The steps of this phase are shown in Figure 5. For a new instance, step 7 (Characteristics
Measurement) calculates its characteristic values using indices. Step 8 uses the learned
classifiers to determine, from the characteristics of the new instance, which group it
belongs to. The algorithm associated to this group is the expected best algorithm for the
instance.
Fig. 5. Steps of the prediction phase
Search WWH ::




Custom Search