Graphics Reference
In-Depth Information
Table 8.11
Wilcoxon test results over medium data sets
tst Acc.
tst Kap.
Acc. * Red.
Kappa. * Red.
+
±
+
±
+
±
+
±
AllKNN
9 9
10 19
34
36
CCIS
17
04
11 18
4 4
CHC
5
20
5 9
19 20
15
20
CNN
4 0
5 3
6 1
7 5
CPruner
3 4
05
8 4
2 5
DROP3
2 1
2 0
9 4
8 5
FCNN
4 2
5 4
6 6
9 7
GGA
5 3
4 4
12 18
12 17
HMNEI
10 19
12
20
5 0
5 4
IB3
2 1
49
9 7
9 6
ICF
04
07
6 1
3 1
MCNN
02
04
10 17
7 8
MENN
11 19
8 8
48
49
ModelCS
10
20
12
20
11
03
MSS
6 8
9 8
3 0
4 1
POP
7
20
10
20
00
02
Reconsistent
1 0
3 0
39
4 1
RMHC
11 19
9 9
13 18
14 19
RNG
15 20
15 20
23
04
RNN
4 4
4 2
14 18
15 19
SSMA
8
20
9 9
19 20
19 20
•
For the tradeoff reduction-accuracy rate: The algorithms which obtain the best
behavior are RMHC and SSMA. However, these methods achieve a significant
improvement in the accuracy rate due to a high computation cost. The methods
that harm the accuracy at the expense of a great reduction of time complexity are
DROP3 and CCIS.
•
If the interest is the accuracy rate: In this case, the best results are to be achieved
with the RNGE as editor and HMNEI as hybrid method.
•
When the key factor is the condensation: FCNN is the highlighted one, being one
of the fastest.
8.6.4 Visualization of Data Subsets: A Case Study Based
on the Banana Data Set
This section is devoted to illustrating the subsets selected resulting from some PS
algorithms considered in our study. To do this, we focus on the banana data set, which
