Spatial Clustering

The only spatial data provided with dataset is the latitude and longitude of venues.

We tried to find higher level meaningful geographical segmentation for Venues which leads us to Spatial data Clustering.

“activity zone” is a term which we assign to each cluster.

As an experiment we ran the 2 clustering algorithms; k-means and DBSCAN. As a conclusion we decided to use both clustering algorithms. DBSCAN for finding first level dense areas then applying k-means on clusters from the first step which are big and could be divided down into smaller regions.