Logical Analysis of Data

Abstract

The logical analysis of data (LAD) is a methodology aimed at extracting or discovering knowledge from data in logical form. The first paper in this area was published as Crama, Hammer, & Ibaraki (1988) and precedes most of the data mining papers appearing in the 1990s. Its primary target is a set of binary data belonging to two classes for which a Boolean function that classifies the data into two classes is built. In other words, the extracted knowledge is embodied as a Boolean function, which then will be used to classify unknown data. As Boolean functions that classify the given data into two classes are not unique, there are various methodologies investigated in LAD to obtain compact and meaningful functions. As will be mentioned later, numerical and categorical data also can be handled, and more than two classes can be represented by combining more than one Boolean function.