Trajectoiries mining : Get_Move, Closed-Multi-Gradual et Real Gpatterns
Get_Move is a prototype which efficiently extracts different types of patterns such as "trajectories" (see Figure 1).
The prototype allows for example to extract the well known patterns: closed-swarm, convoys or group patterns. The originality of Get_Move is that the mining of all these patterns is done simultaneously. The algorithm is based on an array of clusters on which an itemsets mining algorithm is applied. Different properties are used to generate complex trajectories (see Figure 2).
Experiments conducted on real sets of synthetic data showed that Get_Move was often more effective than the specific approaches and that in all cases the extraction is faster than the application of any specific algorithms.
Figure 2 shows examples of patterns extracted with Get_Move method.
Get_Move has been published in IDA 2012 and PKDD 2012. An online démonstration is available.
Last update on 20/01/2015