Parallel construction of decision trees with consistently non-increasing expected number of tests

摘要

In recent years, with the emergence of big data and online Internet applications, the ability to classify huge amounts of objects in a short time has become extremely important. Such a challenge can be achieved by constructing decision trees (DTs) with a low expected number of tests (ENT). We address this challenge by proposing the save favorable general optimal testing algorithm' (SF-GOTA) that guarantees, unlike conventional look-ahead DT algorithms, the construction of DTs with monotonic non-increasing ENT. The proposed algorithm has a lower complexity in comparison to conventional look-ahead algorithms. It can utilize parallel processing to reduce the execution time when needed. Several numerical studies exemplify how the proposed SF-GOTA generates efficient DTs faster than standard look-ahead algorithms, while converging to a DT with a minimum ENT.