Classifying Vehicle Activity to Improve Point of Interest Extraction

Algorithm 4

Activity-based vehicle PoI extraction (AVPE)—deployment stage.

	inputs: , a set of trajectories in the buffer,
	, the merge threshold
	, the pre-trained classifier
	, the feature set used in the classifier
	, the chosen clustering algorithm, with pretrained parameters
	output: , a set of clusters that are considered to be relevant
(1)
(2)	for do
	//calculate features for the cluster from the vehicle signal values, features () is defined in Algorithm 2
	//select feature values from the feature set and obtain prediction from classifier
(3)
	//if the prediction is in the set of positive labels, add the cluster to our return set
(4)	if then
(5)
(6)	end
(7)	end
	//return a set of clusters that are considered to be relevant
(8)	return