72
C.2. Case B: splitting the KDDTrain+ for training and test sets
Logistic Regression:
Table 41: logistic regression on the split multiclass training set
precision recall f1-score support
back
0.99
0.95
0.97
790
buffer_overflow 0.70
0.82
0.76
17
ftp_write
0.17
0.50
0.25
2
guess_passwd
0.97
0.93
0.95
42
imap
0.38
1.00
0.55
3
ipsweep
0.97
0.97
0.97
2924
land
1.00
0.70
0.82
20
loadmodule
0.11
0.25
0.15
4
multihop
0.00
0.00
0.00
1
neptune
1.00
1.00
1.00
32966
nmap
0.95
0.92
0.94
1246
normal
0.99
0.99
0.99
53988
perl
0.00
0.00
0.00
0
phf
1.00
0.57
0.73
7
pod
0.99
1.00
0.99
158
portsweep
0.98
1.00
0.99
2280
rootkit
0.17
0.33
0.22
3
satan
0.94
0.98
0.96
2780
smurf
1.00
0.99
0.99
2164
spy
0.00
0.00
0.00
0
teardrop
1.00
1.00
1.00
731
warezclient
0.80
0.88
0.84
640
warezmaster
0.58
0.58
0.58
12
accuracy
0.99
100778
macro
0.68
0.71
0.68
100778
weighted
0.99
0.99
0.99
100778