Skip to main content

Table 3 Summary statistics for detecting good and bad objects in datasets 7-10 from SCOD algorithm

From: Thresher: determining the number of clusters while removing outliers

Scenarios and datasets 96 variables, 24 objects 24 variables, 96 objects
  Dataset 7 Dataset 8 Dataset 9 Dataset 10 Dataset 7 Dataset 8 Dataset 9 Dataset 10
Sensitivity 0.337 0.344 0.327 0.328 0.225 0.228 0.223 0.217
Specificity 0.670 0.661 0.674 0.658 0.780 0.774 0.780 0.786
FDR 0.660 0.663 0.333 0.342 0.661 0.666 0.333 0.338
AUC 0.504 0.502 0.501 0.493 0.502 0.501 0.502 0.501