Ομαδοποίηση ημερολογιών κίνησης.
Clustering in travel diaries
SubjectΓεωγραφικά Συστήματα Πληροφοριών -- Στατιστικές μέθοδοι ; Στατιστική -- Οικονομετρικές μέθοδοι
The subject of this thesis is to investigate possible interactions between variables which are listed in trajectories of moving objects. The data that were used came from a city in Belgium and were collected from two sources, the «data GPS» which are GPS records per user and more specifically from the GPS enabled devices and the «Diaries» in which each user records the semantic information about the trip. More specifically, in order to achieve our purpose we applied to our data the method of clustering. We applied the stochastic hierarchical clustering algorithm cumulatively by grouping our data in compact clusters ie based on the maximum square Euclidean distance that these clusters have between them. Three different pairs of clustering were tested and the results that we have reached, indicated that only one pair of these variables correlate quite well.