Theory of actionable data mining with application to semiconductor manufacturing control

Dan Braha, Yuval Elovici, Mark Last

International Journal of Production Research 45 (13), 3059-3084, 2007

Accurate and timely prediction of a manufacturing process yield and flow times is often desired as a means of reducing overall production costs. To this end, this paper develops a new decision-theoretic classification framework and applies it to a real-world semiconductor wafer manufacturing line that suffers from constant variations in the characteristics of the chip-manufacturing process. The decision-theoretic framework is based on a model for evaluating classifiers in terms of their value in decision-making. Recognizing that in many practical applications the values of the class probabilities as well as payoffs are neither static nor known exactly, a precise condition under which one classifier ‘dominates’ another classifier (i.e. achieves higher payoff), regardless of payoff or class distribution information, is presented. Building on the decision-theoretic model, two robust ensemble classification methods are proposed …