METHOD FOR THE STATISTICAL CLASSIFICATION OF DEBTOR CLIENTS FOR ANALYSING PAYMENT PROBABILITY

摘要

The invention relates to a method for the evolutionary identification of debtor client classification attributes that can predict payment probability, and for establishing different efficient client attention strategies for clients with different profiles. The method uses decision trees for identifying attributes that have a higher weighting in a client's payment probability distribution, carries out a second estimation of the same probability by means of Bayesian networks, and combines the two estimations to obtain a prediction with a high level of certainty.