Thesis APLICACION DE SEGMENTACION DE CLIENTES MOROSOS PARA UNA EMPRESA DE COBRANZAS
Loading...
Date
2013-05
Journal Title
Journal ISSN
Volume Title
Program
DEPARTAMENTO DE INGENIERÍA COMERCIAL. MAGÍSTER EN GESTIÓN EMPRESARIAL-MBA
Campus
Casa Central Valparaíso
Abstract
El trabajo que se muestra a continuación presenta una metodología que permite obtener
una segmentación para la cartera de clientes morosos de una empresa de cobranzas del
sector Financiero en Chile. Para lograr obtener esta segmentación, se hace uso de la
metodología que propone el proceso KDD (Knowledge Discovery in Databases). Este
proceso estructurado permite mediante una serie de etapas obtener conocimiento desde
los datos que se encuentran disponibles en los sistemas de Información de la Empresa de
Cobranzas.
La segmentación de la cartera de clientes morosos se elabora con la finalidad de generar
estrategias adecuadas aplicables a cada segmento de clientes. Esta tarea trae como
beneficio principal el conseguir un mayor grado de eficiencia en los recursos y
herramientas de gestión que se utilizan para desarrollar la Cobranza, dado que permite
obtener conocimientos útiles para el desarrollo de nuevas estrategias de gestión.
Para la generación de los modelos se utilizan 2 Algoritmos de segmentación: Two-step o
segmentación en 2 etapas y el Algoritmo K-means.
Al comparar ambos resultados se aprecian muchas similitudes en las características de la
composición de los 5 segmentos que entrega como resultado cada modelo, sin embargo,
se escoge el resultado entregado por el modelo k-means. El análisis detallado de los
segmentos obtenidos, permite clasificarlos en base a sus características y renombrarlos
según sus principales atributos. De esta manera los segmentos obtenidos son: Inubicable
Temprano (Poca Morosidad, pero no contactable), Moroso Responsable (está moroso,
pero tiene buen comportamiento de pago), Década Pasada (créditos más antiguos
vendidos antes del 2010), Renegociado nuevo (Créditos renegociados cursados desde el
2011) y Mora Dura (Cliente con mayor cantidad de cuotas morosas en promedio).
The following work presents a methodology to obtain segmentation for delinquent customer base for a collection company in financial sector in Chile. In order to obtain this segmentation, it makes use of the proposed methodology for the KDD process (Knowledge Discovery in Databases). This structured process allows through a series of steps to gain knowledge from the data available in the Information systems of the Collection Company. The segmentation of the delinquency customer base is made in order to generate appropriate strategies applicable to each customer segment. This task brings the primary benefit of achieving a greater degree of efficiency in the management resources and tools that are used to make the Collection, as it allows to obtain useful knowledge for the development of new management strategies. Generation models use two segmentation algorithms: Two-step algorithm and K-means algorithm. Comparing both results were found many similarities in the characteristics of the composition of the 5 segments each model delivers results, however, the result is selected by the model given k-means. Detailed analysis of the segments obtained, can be classified based on their characteristics and name them according to their main attributes. In this way the segments obtained are: Early unreachable (Little Delinquency, but lost), Responsible Delinquent (is delinquent, but has good payment), Past Decade (oldest credits sold before 2010), New Rewrite (Credits renegotiated sold since 2011), Hard Delinquent (Client with the most delinquent fees on average).
The following work presents a methodology to obtain segmentation for delinquent customer base for a collection company in financial sector in Chile. In order to obtain this segmentation, it makes use of the proposed methodology for the KDD process (Knowledge Discovery in Databases). This structured process allows through a series of steps to gain knowledge from the data available in the Information systems of the Collection Company. The segmentation of the delinquency customer base is made in order to generate appropriate strategies applicable to each customer segment. This task brings the primary benefit of achieving a greater degree of efficiency in the management resources and tools that are used to make the Collection, as it allows to obtain useful knowledge for the development of new management strategies. Generation models use two segmentation algorithms: Two-step algorithm and K-means algorithm. Comparing both results were found many similarities in the characteristics of the composition of the 5 segments each model delivers results, however, the result is selected by the model given k-means. Detailed analysis of the segments obtained, can be classified based on their characteristics and name them according to their main attributes. In this way the segments obtained are: Early unreachable (Little Delinquency, but lost), Responsible Delinquent (is delinquent, but has good payment), Past Decade (oldest credits sold before 2010), New Rewrite (Credits renegotiated sold since 2011), Hard Delinquent (Client with the most delinquent fees on average).
Description
Keywords
SEGMENTACIÓN DE CARTERA