Discretization of Rational Data

Frequently one wants to extend the use of a classification method that in principle requires records with True/False values, so that records with rational numbers can be processed. In such cases, the rational numbers must first be replaced by True/False values before the method may be applied. In other cases, a classification method in principle can process records with rational numbers directly, but replacement by True/False values improves the performance of the method. The replacement process is usually called discretization or binarization. This paper describes a recursive discretization process called Cutpoint. The key step of Cutpoint detects points where classification patterns change abruptly. The paper includes computational results where Cutpoint is compared with entropy-based methods, which to-date have been found to be the best discretization schemes. The results indicate that Cutpoint is preferred by certain classification schemes, while entropy-based methods are better for other classification methods. Thus, one may view Cutpoint to be an additional discretization tool that one may want to consider.