Example:
Scalable Tree Algorithms
Scaling decision tree generation to large databases
Would like to:
remove requirement that data fit in memory
optimize data scans
e.g. SLIQ: Mehta, Agrawal, and Rissanen (1996)
