Discovering Travel Spatiotemporal Pattern Based on Sequential Events Similarity

Algorithm 4

Travel sequence clustering algorithm (TSCA).

	Input: travel sequences set TS = {L1, L2, …, Ln} and the number of clusters k
	Output: travel sequence cluster set TC = {TC1, TC2, …, TCk} and k center sequences set newMedoids = {L1, L2, …, LK}
	Initialization: oldMedoids ⟵ null, newMedoids ⟵ null;
(1)	Select k sequences L1, L2, …, Lk from TS randomly as initial center sequences to oldMedoids;
(2)	TCi ⟵ Li //Each center sequence corresponds to a cluster
(3)	while (!isEqual (oldMedoids, newMedoids))
(4)	Calculate the similarity of each sample sequence from TS to each center sequence from newMedoids and place the sample sequence in the cluster with the highest similarity to the center sequence;
(5)	oldMedoids ⟵ newMedoids;
(6)	Recalculate the center sequence of each cluster TCi, sequences with the highest similarity from each sample sequence in the cluster, as newMedoids;
(7)	return TC and newMedoids;