Table 5

Performance measures of data mining algorithm at different levels of significance over Asthma dataset 4 classes
SIGNIFICANCE p < 5 x 10-5 p < 5 x 10-4 p < 5 x 10-3 p < 5 x 10-2
Algorithm
    Acc.
    Sp
    Sn
    AUC
    Acc.
    Sp
    Sn
    AUC
    Acc.
    Sp
    Sn
    AUC
    Acc
    Sp
    Sn
    AUC
    Avg.
Naïve Bayes 61.7 87.2 61.7 0.82 68.1 89.3 68.1 0.86 72.3 90.8 72.3 0.87 70.2 90.0 70.2 0.86 77.7
SLR 57.5 85.8 57.4 0.80 57.4 85.6 57.4 0.81 72.3 90.7 72.3 0.85 55.3 86.1 55.3 0.76 72.2
SVM 55.3 86.2 55.3 0.77 55.3 86.2 55.3 0.77 61.7 87.2 61.7 0.82 66.0 87.6 66.0 0.81 71.3
MLP 55.3 86.1 55.3 0.82 53.2 84.6 53.2 0.80 63.8 87.8 63.8 0.88 dnf dnf dnf dnf 71.1*
Logistic R. 48.9 87.0 48.9 0.78 53.2 84.4 53.2 0.79 59.6 86.4 59.6 0.84 68.0 89.2 68.1 0.86 70.8
R. Forest 48.9 86.9 48.9 0.77 48.9 86.9 48.9 0.77 46.8 81.1 46.8 0.75 40.4 80.0 40.4 0.71 62.8
VFI 48.9 82.8 48.9 0.66 48.9 82.9 48.9 0.67 51.0 83.6 51.1 0.69 46.8 81.9 46.8 0.77 62.6
Hyper Pipes 51.1 83.4 51.1 0.72 53.2 84.0 53.2 0.70 46.8 71.8 46.8 0.74 42.6 80.3 42.0 0.75 62.3
M5P 48.9 82.8 48.9 0.79 55.3 86.1 55.3 0.81 42.5 81.0 42.6 0.68 27.6 75.8 27.7 0.57 60.0
KNN 42.5 87.1 42.6 0.69 46.8 86.6 46.8 0.67 44.6 88.0 44.7 0.69 36.2 79.7 36.2 0.67 59.6
K means 40.4 81.9 40.4 0.60 46.8 82.2 46.8 0.65 42.6 80.7 42.6 0.62 34.0 78.0 34.0 0.56 55.8
Bayes Net 38.3 79.3 38.3 0.56 36.2 77.8 36.2 0.56 44.7 81.4 44.7 0.63 36.2 77.6 36.2 0.60 53.9
K star 48.9 83.0 48.9 0.70 38.3 79.4 38.3 0.63 36.2 79.4 36.2 0.62 23.4 76.4 23.4 0.49 53.5
Random Tree 29.8 76.6 29.8 0.53 40.4 80.2 40.4 0.60 38.3 79.5 38.3 0.59 40.4 80.2 40.4 0.60 52.9
LDA 53.2 84.4 53.2 0.80 27.7 80.0 32.5 0.57 8.5 86.5 16.7 0.56 14.9 83.6 23.3 0.53 50.7
J48 27.7 75.4 27.7 0.52 27.7 75.9 27.7 0.49 42.6 80.8 42.6 0.58 31.9 77.1 31.9 0.52 48.7
ASC 27.7 76.0 27.7 0.52 19.2 71.8 19.1 0.46 29.8 76.7 29.8 0.52 21.2 74.8 21.3 0.45 43.1

Acc: Accuracy, Sp: Specificity, Sn: Sensitivity, AUC: Area under ROC curve, Avg: Average score in % for each algorithms, dnf: Did not Finish”, * denotes Avg. from 3 significance levels. Measures >90% are marked in bold.

Kukreja et al.

Kukreja et al. BMC Bioinformatics 2012 13:139   doi:10.1186/1471-2105-13-139

Open Data