Revenue recovering with insolvency prevention on a Brazilian telecom operator

  • Authors:
  • Carlos André R. Pinheiro;Alexandre G. Evsukoff;Nelson F. F. Ebecken

  • Affiliations:
  • Brasil Telecom, Brasília, DF, Brazil;COPPE/UFRJ, Rio de Janeiro, RJ, Brazil;COPPE/UFRJ, Rio de Janeiro, RJ, Brazil

  • Venue:
  • ACM SIGKDD Explorations Newsletter
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper deals with a real application on a brazilian telephone company. Data mining analysis, based on neural networks, was performed on the customer base in order to understand and to prevent bad debt events. This paper describes the knowledge discovering process and focuses on two main products: the cluster analysis of the customer base and a bad debt event classification model. The segmentation of the customer base has provided a better understanding of several groups of customers' behavior. Distinct actions are taken, depending on the segment a given client was put in, according to strategic directions of the company. The classification of insolvent customers is used as a tool to help the company to take preventing actions in order to avoid main losses and taxes leakage. The results of the project's implantation show that investment on information technology infra-structure for data mining is highly profitable.