Investigating Tree Family Machine Learning Techniques for a Predictive System to Unveil Software Defects
Table 19
Results of different TF-ML techniques in terms of accuracy along with the rank values.
Datasets
CDT
CS-forest
DS
Forest-PA
HT
J48
LMT
RF
RT
REP-T
AR1
92.56 (1.3)
85.12 (6)
90.90 (3)
91.73 (2.5)
92.56 (1.3)
90.08 (4.5)
91.73 (2.5)
90.08 (4.5)
89.25 (5)
92.56 (1.3)
AR3
87.30 (4.2)
84.12 (5.2)
90.4762 (2)
88.88 (3)
84.12 (5.2)
87.30 (4.2)
87.30 (4.2)
92.06 (1)
87.30 (4.2)
87.30 (4.2)
CM1
89.35 (3)
82.53 (7)
90.16 (1.2)
89.95 (2)
90.16 (1.2)
87.95 (5)
89.15 (4.3)
89.15 (4.3)
83.33 (6)
89.15 (4.3)
KC2
82.95 (4)
79.11 (9)
79.69 (8)
83.52 (2)
83.33 (3.5)
81.41 (6)
84.29 (1)
83.33 (3.5)
80.84 (7)
81.60 (5)
KC3
81.95 (1.2)
81.44 (2.5)
81.95 (1.2)
79.89 (4)
80.92 (3)
79.38 (5.3)
79.38 (5.3)
81.44 (2.5)
70.61 (6)
79.38 (5.3)
MW1
92.30 (2.5)
88.33 (6)
91.06 (5)
92.05 (3.3)
92.30 (2.5)
92.05 (3.3)
93.05 (1)
92.05 (3.3)
86.60 (7)
91.56 (4)
PC1
93.50 (3.5)
91.16 (7)
93.05 (5.2)
93.50 (3.5)
93.05 (5.2)
93.32 (4)
92.42 (6)
93.68 (1)
91.07 (8)
93.59 (2)
PC2
99.58 (1.14)
99.51 (3)
99.58 (1.14)
99.58 (1.14)
99.58 (1.14)
99.58 (1.14)
99.57 (2)
99.58 (1.14)
99.19 (4)
99.58 (1.14)
PC3
89.18 (5)
84.77 (9)
89.76 (2.5)
89.69 (3)
89.76 (2.5)
88.93 (7)
89.12 (6)
90.14 (1)
85.54 (8)
89.63 (4)
PC4
89.16 (6)
88.88 (7)
87.79 (9)
90.05 (3)
88.06 (8)
89.36 (5)
90.32 (2)
90.67 (1)
86.69 (10)
89.71 (4)
Sum (rank)
31.84
61.7
38.24
27.44
33.54
45.44
34.3
23.24
65.2
35.24
Average (rank)
3.18
6.17
3.82
2.74
3.35
4.54
3.43
2.32
6.52
3.52
Sum (Acc)
897.88
865.02
894.46
898.91
893.89
889.41
896.36
902.23
860.45
894.10
Average (Acc)
89.78
86.50
89.44
89.89
89.38
88.94
89.63
90.22
86.04
89.41
It ranks the technique for each data set separately, the best performing algorithm getting the rank of 1 and the second-best rank 2. Last two columns present the sum and average of ranks for each technique.
