Data mining is useful in making single decisions. The problem is when there are several related problems and the best local decisions do not make the best global result. We propose to calibrate each local data mining ...