Increasing the Accuracy of Software Fault Prediction Using Majority Ranking Fuzzy Clustering* - Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing - page 183

We consider FMR row
posed model performed
performed well in terms
FNR rate in ACN were t
ACN was used. Error rat
JM1 and AR5.As it is als
outperformed other meth
and AR5. According to
outperformed other metho
test cases, so we accept hy

w in Tables 1 to 7 to answer research question 2. Our pro
best in all evaluation criteria in PC1 and AR4. It als
of FNR in all datasets except KC2 and CM1, which th
the best. FPR rate in KC1 and JM1 were the best whe
te was also the best in our proposed method except fo
so shown in Figs. 3 and 4, our proposed model (FMR)
hods in terms of overall error in all datasets except JM
the results, we can conclude that our proposed metho
od in terms of error and FNR ratein more than half of th
ypothesis 2.

o-
so
he
en
or
is
M1
od
he

0.5

0.45

0.4

0.35

0.3

0.25

0.2

0.15

0.1

0.05

0

KC1 KC2 CM1 PC1 JM1 AR4 AR5

Fig. 3 Method comparisons b based on false negative rate (FNR)

0.12

0.1

0.08

0.06

0.04

0.02

0

KC1 KC2 CM1 PC1 JM1 AR4 AR5

Fig. 4 Method comparisons b based on overall error rate