Information Technology Reference
In-Depth Information
For this problem, the function set F = {+, -, *, LT, GT, LOE, GOE, ET,
NET} (the last six functions are comparison functions of two arguments
which return 1 if the condition is true or 0 if false, representing, respectively,
less than, greater than, less or equal to, greater or equal to, equal to, and not
equal to), in which each function was weighted 5 times; and the set of termi-
nals included all the 51 attributes which were represented by
d
0
-
d
50
. The 0/1
rounding threshold
R
was equal to 0.5 and the fitness was based on the number
of hits and was evaluated by equation (3.8). Thus, in this case,
f
max
= 345.
For this problem, chromosomes composed of 10 genes with an
h
= 7 and
sub-ETs linked by addition were used. The parameters used per run are sum-
marized in Table 4.6. Note again that, despite the high dimensionality of the
problem, small populations of just 30 individuals were used, as this allows a
quick and efficient evolution.
In one run, by generation 11446 a very good solution with a classification
error of 10.145% and a classification accuracy of 89.855% was created (the
Table 4.6
Settings used in the credit screening problem.
Population size
30
Number of training samples
345
Number of testing samples
172
Function set
(+ - * LT GT LOE GOE ET NET)
5
Terminal set
d0 - d50
Head length
7
Gene size
15
Number of genes
10
Linking function
+
Chromosome length
150
Mutation rate
0.044
Inversion rate
0.1
IS transposition rate
0.1
RIS transposition rate
0.1
Gene transposition rate
0.1
One-point recombination rate
0.3
Two-point recombination rate
0.3
Gene recombination rate
0.3
Fitness function
Equation (3.8)
Rounding threshold
0.5
