An Efficient Outlier Detection Approach for Streaming Sensor Data Based on Neighbor Difference and Clustering

Table 3

The comparisons of detection accuracy

Types	Variables
	ODNDC (%)	LODA	ROCF	CORM	Heuristic (%)
Point outliers	90.9	—	—	—	86.4
Collective outliers	94.4	—	—	—	92.5
Jump outliers	95.8	—	—	—	87.5
Contextual outliers	94.2	—	—	—	90.7
Overall	94.3	93.6%	92.8%	89.5%	92