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

Algorithm 2

Outlier Detection based on Clustering (ODC).

	Input: the data labeled by ODND
	Output: outliers and their types
(01)	use a sliding window model to receive data from all sensors
(02)	for each point outlier detected by ODND do
(03)	replace them by their mean values calculated with adjacent data instances
(04)	end for
(05)	detect outliers using w-k-means algorithm
(06)	Reassign the labels of all outliers detected by ODND and w-k-means algorithm