Metacost: a General Method for Making Classiiers Cost-sensitive

Research in machine learning, statistics and related elds has produced a wide variety of algorithms for classiication. However, most of these algorithms assume that all errors have the same cost, which is seldom the case in KDD problems. Individually making each classiication learner cost-sensitive is laborious, and often non-trivial. In this paper we propose a principled method for making an arbitrary classi-er cost-sensitive by wrapping a cost-minimizing procedure around it. This procedure, called MetaCost, treats the underlying classiier as a black box, requiring no knowledge of its functioning or change to it. Unlike stratiication, Meta-Cost is applicable to any number of classes and to arbitrary cost matrices. Empirical trials on a large suite of benchmark databases show that MetaCost almost always produces large cost reductions compared to the cost-blind classiier used (C4.5RULES) and to two forms of stratiication. Further tests identify the key components of MetaCost and those that can be varied without substantial loss. Experiments on a larger database indicate that MetaCost scales well.