machine learning - DBSCAN using spatial and temporal data -
i looking @ data points have lat, lng, , date/time of event. 1 of algorithms came across when looking @ clustering algorithms dbscan. while works ok @ clustering lat , lng, concern fall apart when incorporating temporal information, since it's not of same scale or same type of distance.
what options incorporating temporal data dbscan algorithm?
look generalized dbscan same authors.
sander, jörg; ester, martin; kriegel, hans-peter; xu, xiaowei (1998). density-based clustering in spatial databases: algorithm gdbscan , applications. data mining , knowledge discovery (berlin: springer-verlag) 2(2): 169–194. doi:10.1023/a:1009745219419.
for (generalized) dbscan, need 2 functions:
findneighbors - "related" objects database
corepoint - decide whether set enough start cluster
then can repeatedly find neighbors grow clusters.
function 1 want hook into, example using 2 thresholds: 1 geographic , 1 temporal (i.e. within 100 miles, , within 1 hour).
Comments
Post a Comment