Table 11.
Average area under the ROC curve (AUC) on unsupervised anomaly detection tasks. The highest average AUC score among the anomaly detection methods is shown in bold. When FRaC has the highest average AUC score, we show the p-value of a paired, one-tailed t-test comparing the AUC scores of FRaC to the next best anomaly detection method. p-values under 0.05 are shown in bold
| Data set | One-class SVM | LOF | CFA | FRaC | p-value |
|---|---|---|---|---|---|
| abalone | 0.56 | 0.48 | 0.42 | 0.46 | |
| acute | 0.94 | 0.83 | 1.00 | 0.98 | |
| adult | 0.48 | 0.47 | 0.52 | 0.59 | 9.86e – 15 |
| annealing | 0.63 | 0.83 | 0.77 | 0.79 | |
| arrhythmia | 0.44 | 0.75 | 0.43 | 0.80 | 0.00263 |
| audiology | 0.70 | 0.72 | 0.75 | 0.77 | 0.0241 |
| balance-scale | 0.95 | 0.96 | 0.95 | 0.97 | 0.00289 |
| blood-transfusion | 0.51 | 0.56 | 0.45 | 0.56 | |
| breast-cancer-wisconsin | 0.51 | 0.92 | 0.28 | 0.96 | 1.7e – 11 |
| car | 0.98 | 1.00 | 0.97 | 0.96 | |
| chess | 0.70 | 0.89 | 0.90 | 0.90 | |
| cmc | 0.48 | 0.47 | 0.41 | 0.43 | |
| connect-4 | 0.52 | 0.86 | 0.70 | 0.63 | |
| credit-screening | 0.51 | 0.69 | 0.78 | 0.84 | 0.00252 |
| cylinder-bands | 0.37 | 0.71 | 0.68 | 0.61 | |
| dermatology | 0.95 | 0.98 | 0.74 | 0.99 | 0.063 |
| echocardiogram | 0.54 | 0.67 | 0.51 | 0.71 | 0.0138 |
| ecoli | 0.98 | 0.98 | 0.34 | 0.96 | |
| glass | 0.64 | 0.75 | 0.50 | 0.65 | |
| haberman | 0.60 | 0.67 | 0.70 | 0.70 | |
| hayes-roth | 0.94 | 0.67 | 0.91 | 0.94 | |
| hepatitis | 0.51 | 0.42 | 0.74 | 0.83 | 0.0187 |
| horse-colic | 0.64 | 0.62 | 0.56 | 0.78 | 4.01e – 06 |
| image | 0.55 | 0.93 | 0.31 | 0.97 | 0.0316 |
| internet_ads | 0.77 | 0.78 | 0.89 | 0.91 | 0.000689 |
| ionosphere | 0.84 | 0.90 | 0.50 | 0.96 | 8.85e-08 |
| iris | 1.00 | 1.00 | 0.58 | 1.00 | |
| letter-recognition | 0.99 | 0.99 | 0.91 | 1.00 | 0.00316 |
| libras | 0.57 | 0.76 | 0.56 | 0.86 | 7.8e – 06 |
| magic | 0.50 | 0.81 | 0.57 | 0.81 | |
| mammographic-masses | 0.74 | 0.57 | 0.67 | 0.71 | |
| mushroom | 0.94 | 0.99 | 0.79 | 0.91 | |
| nursery | 1.00 | 1.00 | 1.00 | 1.00 | |
| ozone | 0.49 | 0.45 | 0.33 | 0.30 | |
| page-blocks | 0.49 | 0.93 | 0.37 | 0.81 | |
| parkinsons | 0.52 | 0.66 | 0.31 | 0.56 | |
| pima–indians-diabetes | 0.52 | 0.71 | 0.47 | 0.75 | 7.89e – 07 |
| poker | 0.53 | 0.52 | 0.54 | 0.56 | 0.00017 |
| secom | 0.50 | 0.53 | 0.52 | 0.54 | 0.064 |
| spambase | 0.77 | 0.58 | 0.77 | 0.80 | 0.0308 |
| statlog | 0.50 | 0.57 | 0.56 | 0.62 | 0.0112 |
| tae | 0.44 | 0.37 | 0.55 | 0.58 | 0.260 |
| tic-tac-toe | 0.83 | 1.00 | 1.00 | 0.99 | |
| voting-records | 0.98 | 0.69 | 0.60 | 0.87 | |
| wine | 0.48 | 0.87 | 0.11 | 0.94 | 1e – 05 |
| yeast | 0.66 | 0.71 | 0.51 | 0.70 | |
| zoo | 0.99 | 1.00 | 1.00 | 1.00 | |
| Number of data sets with the maximum AUC score | 5 | 9 | 1 | 22 | |