Search

The purpose of the article is to demonstrate how entropy and some theorems of the theory of information may be useful in the domain of classification, both for class formation and for class identification. Starting from a sample of subjects who have given different answers to a battery of tests or items, researchers see that there can be two kinds of approaches. Firstly, on the basis of entropy one can sort out the tests that have higher information content and on this basis form classes of subjects whose homogeneity can be appraised. In this case the problem is therefore to form a certain number of homogeneous classes with the minimum number of tests. Secondly, once the classification of subjects has been made, it is possible, according to the theory of information, to see whether the typical sequence of subjects of the different classes can be established also through fewer tests, being it then possible also to recognize a distance between the different classes. With this method researchers try to arrive at homogeneous classes on the basis of the smallest possible number of tests, that is in effect choosing variables with the greatest discriminating power.