A Method for Controlling

1 Abstract Two types of errors can be associated with classifying data into two categories A and B: misclassifying an item in A as one of B and, conversely, misclassifying an item in B as one of A. Tight control over these two errors is needed in many practical situations. For example, an error of one type may have far more serious consequences than one of the other type. Most previous work on two-class classiiers does not allow for such control. In this paper, we describe a general approach that supports tight error control for two-class classiication and that can utilize almost any two-class classiication method as part of the decision mechanism. The main idea is to construct from the given training data a family of classiiers and then, using the training data once more, to estimate two distributions of certain vote totals. The error control is achieved via the two estimated distributions. The control is eeective if the estimates of the two distributions are close to the true distributions. The approach has been tested using six well-known classiication problems. In each case, the two estimated distributions were close or very close to those obtained from veriication data. Accordingly, error control would be good. In addition, when the goal was minimization of total errors, then the accuracy was essentially as good as that of the best prior methods.