| Tech | Version |
|---|---|
| Scala | 2.11.7 |
| java | 1.8 |
| Hadoop | 2.7.2 |
| HBase | 1.2.1 |
- Input: Data in the form (id, type, (latitude, longitude))
- Map data points to different grids.
- Use plane-sweep algorithm to find neighbors for each data point.
- Perform neighbor grouping.
- Count instances for different types.
- Generate size k co-locations.
HBase is used at following places:
-
Save (event, count) pairs in reducer for counting instances of different event types. Here, we can use event as the row key and count as the value.
-
Save prevalent colocation patterns in reducer for co-location pattern search. Here we can use the eventset as the row key, size as the column key and [instance] as the value.
-
Read size k-1 colocations in scanNTransactions method in mapper for co-location pattern search. Here, the lookup can be performed easily using the row key for a given size (column key).
- The algorithm does not perform candidate set generation.