WIT Press

Knowledge Discovery For CRM Improvement


Free (open access)

Paper DOI






Page Range

175 - 184




588 kb


G. M. Caputo, V. M. Bastos, A. M. Cister & N. F. F. Ebecken


The objective of this paper is to present a database marketing analysis through data and text mining tools. A case study of a Brazilian Power Energy distribution was developed indoors. The main idea is to transform the database information into strategic marketing knowledge. Thus a data warehouse sample was treated, reduced and clustered. Principal component analysis was used to reduce the original number of variables. The entire database was classified after creation by the decision trees and neural networks approach. In this work, text mining techniques were used to process customers’ claims in order to improve cluster results. The CRM group has developed a powerful tool to gather knowledge regarding the skills and habits of customers, thereby gaining their confidence and loyalty. Keywords: data mining, text mining, clustering, CRM. 1 Introduction This work shows a methodology to extract knowledge of customer habits from an electric energy company database, in order to discover useful information that will serve as base for CRM implementation in the company. For this, segmentation variables have been used that identify partnereconomic characteristics, behaviors and customer invoicing data, which represent the customer profile for the company. The set of initial variables extracted from the company Data Warehouse involved a great number of dimensions, with many forms of customers’ segment identification. Through the data extraction process, it was possible to get samples with good universal representation, allowing the application of knowledge extraction techniques, in order to identify customers’ behavior.


data mining, text mining, clustering, CRM.