- rebuild() optimization.
- gridsearch(weak-treenode-condition parameter sweep).
- any other ideas?
- survival of the fittest: managing the life time of trees.
- restapi server.
- dynamically adding fields to instances.
- dynamically adding values to label.
- multi-threaded parallel train().
- change the value of any field in an instance (including labels).
- monitoring (with trigger?).
- cache synchronization.
- event log.
- avoid hard-coded train options.
- reducing the size of model in the DB (or limit size).
- support other databases(postgresql, ...).
- tree model to json.
- first release.
- reducing the number of instances in the DB (or limit size).
-
nothing