Table 3

Comparison of classification accuracies of best WEKA classifiers with the MILP based hyper-boxes classification.

% accuracy
% accuracy


ACHE
 7-attribute
 10-attribute
 15-attribute
 BZR
 7-attribute
 10-attribute
 15-attribute
MILP based hyper-boxes method
 100
 91.89
 89.19
 MILP based hyper-boxes method
 96.36
 94.55
 92.73
Bayes Network
 79.28
 77.48
 78.38
 Bayes Network
 77.91
 77.3
 73.62
Naive Bayes
 80.18
 80.18
 81.08
 Naive Bayes
 80.37
 77.91
 66.26
Naive Bayes Simple
 81.08
 80.18
 81.98
 Naive Bayes Simple
 79.14
 77.3
 68.71
Naive Bayes Updatable
 80.18
 80.18
 81.08
 Naive Bayes Updatable
 80.37
 77.91
 66.26
Lojistic
 79.28
 84.68
 80.18
 Lojistic
 83.44
 80.98
 80.98
Multilayer Perceptron
 82.88
 81.08
 81.08
 Multilayer Perceptron
 79.75
 80.98
 79.14
SimpleLogistic
 83.78
 82.88
 79.28
 SimpleLogistic
 80.98
 82.82
 79.14
SMO (WEKA SVM)
 79.28
 80.18
 80.18
 SMO (WEKA SVM)
 79.14
 77.91
 77.91
IB1
 70.27
 80.18
 77.48
 IB1
 72.39
 74.85
 75.46
Ibk
 70.27
 80.18
 77.48
 IBk
 72.39
 74.85
 75.46
Logit Boost
 82.88
 81.08
 82.88
 Logit Boost
 78.53
 77.3
 77.91
Multi Class Classifier
 79.28
 84.68
 80.18
 Multi Class Classifier
 83.44
 80.98
 80.98
Threshold Selector
 47.75
 68.47
 60.36
 Threshold Selector
 78.53
 76.69
 75.46
LMT
 83.78
 82.88
 79.28
 LMT
 80.98
 82.82
 79.14
RandomForest
 80.18
 80.18
 81.98
 RandomForest
 77.3
 79.75
 80.98
OneR
 81.08
 72.97
 72.97
 OneR
 74.85
 74.23
 79.14

% accuracy
% accuracy


DHFR_TG
 7-attribute
 10-attribute
 15-attribute
 COX-2
 7-attribute
 10-attribute
 15-attribute
MILP based hyper-boxes method
 97.74
 96.24
 97.74
 MILP based hyper-boxes method
 98.13
 97.2
 90.65
Bayes Network
 77.33
 78.09
 73.05
 Bayes Network
 67.2
 67.2
 66.88
Naive Bayes
 76.57
 79.35
 72.54
 Naive Bayes
 71.66
 70.06
 64.65
Naive Bayes Simple
 75.57
 78.84
 67
 Naive Bayes Simple
 72.29
 70.06
 64.65
Naive Bayes Updatable
 76.57
 79.35
 72.54
 Naive Bayes Updatable
 71.66
 70.06
 64.65
Lojistic
 75.82
 78.84
 75.57
 Lojistic
 72.29
 70.38
 70.06
Multilayer Perceptron
 76.32
 77.08
 75.06
 Multilayer Perceptron
 72.61
 72.29
 75.16
SimpleLogistic
 74.56
 77.83
 75.31
 SimpleLogistic
 72.29
 71.97
 68.47
SMO (WEKA SVM)
 72.54
 79.09
 72.54
 SMO (WEKA SVM)
 71.02
 69.43
 69.43
IB1
 75.31
 79.09
 75.82
 IB1
 69.11
 71.02
 70.06
Ibk
 75.31
 79.09
 75.82
 IBk
 69.11
 71.02
 70.06
Logit Boost
 77.33
 78.34
 78.34
 Logit Boost
 71.66
 70.06
 70.7
Multi Class Classifier
 75.82
 78.84
 75.57
 Multi Class Classifier
 72.29
 70.38
 70.06
Threshold Selector
 69.77
 74.81
 73.55
 Threshold Selector
 68.47
 65.29
 64.65
LMT
 76.07
 76.57
 77.83
 LMT
 71.34
 71.02
 68.15
RandomForest
 77.58
 79.09
 80.35
 RandomForest
 71.97
 74.2
 70.06
OneR
 69.77
 69.77
 70.53
 OneR
 70.7
 70.38
 70.06

% accuracy
% accuracy


DHFR_RL
 7-attribute
 10-attribute
 15-attribute
 DHFR_PC
 7-attribute
 10-attribute
 15-attribute
MILP based hyper-boxes method
 96.99
 97.74
 94.73
 MILP based hyper-boxes method
 97.62
 98.41
 93.65
Bayes Network
 63.72
 71.78
 70.5
 Bayes Network
 80.42
 80.42
 78.04
Naive Bayes
 63.97
 68.76
 71.7
 Naive Bayes
 82.54
 81.48
 80.95
Naive Bayes Simple
 63.97
 67.75
 71
 Naive Bayes Simple
 82.8
 79.89
 81.22
Naive Bayes Updatable
 63.98
 68.77
 71.78
 Naive Bayes Updatable
 82.54
 81.48
 80.95
Lojistic
 69.52
 73.8
 78.58
 Lojistic
 81.75
 83.33
 81.75
Multilayer Perceptron
 62.72
 76.57
 77.58
 Multilayer Perceptron
 82.8
 82.8
 84.13
SimpleLogistic
 66.75
 73.55
 78.33
 SimpleLogistic
 80.42
 84.13
 81.22
SMO (WEKA SVM)
 64.99
 73.05
 79.59
 SMO (WEKA SVM)
 82.28
 83.33
 79.1
IB1
 62.97
 75.06
 81.11
 IB1
 82.28
 80.16
 81.75
Ibk
 62.97
 75.06
 81.11
 IBk
 82.28
 80.16
 81.75
Logit Boost
 64.99
 75.06
 77.33
 Logit Boost
 83.33
 81.48
 81.48
Multi Class Classifier
 69.52
 73.8
 78.59
 Multi Class Classifier
 81.75
 83.33
 81.75
Threshold Selector
 64.99
 69.52
 78.59
 Threshold Selector
 83.33
 79.1
 81.22
LMT
 65.24
 77.33
 77.83
 LMT
 83.6
 83.07
 85.19
RandomForest
 68.51
 77.08
 77.83
 RandomForest
 82.8
 80.95
 83.07
OneR
 61.46
 66
 62.72
 OneR
 79.89
 79.89
 80.16

Armutlu et al. BMC Bioinformatics 2008 9:411   doi:10.1186/1471-2105-9-411