Classification of Clients on the Basis of Modifying Case-Based Reasoning Algorithms

Abstract

This article deals with client's classification possibilities for trade companies. One of the options for dealing with classification is to use Case-based Reasoning method. The proposed modified algorithm clears cases base, it eliminates invalid data and out-dated data from the time point of view. Hill-climbing algorithm is used for this. Classification results achieved by means of modified algorithm are compared with other classification methods such as Neuron Networks, Top Down Induction Decision Trees and Logistic Regression.