An Improved Semisupervised Outlier Detection Algorithm Based on Adaptive Feature Weighted Clustering

Table 1

Description of real-world datasets.

Dataset	Instances	Features	Outlying classes	Outliers (percent)	Clusters	Prior information
Iris	126	4	“Virginica”	26 (20.63%)	2	10 labeled normal samples, 4 labeled outliers
Abalone	4177	8	“1”–“4,” “16”–“27,” “29”	335 (8.02%)	11	11 labeled normal samples, 18 labeled outliers
Wine	130	13	“3”	11 (8.46%)	2	9 labeled normal samples, 4 labeled outliers
Ecoli	336	9	“omL,” “imL,” “imS”	9 (2.68%)	5	11 labeled normal samples, 3 labeled outliers
WDBC	387	30	“Malignant”	30 (7.75%)	1	10 labeled normal samples, 8 labeled outliers