Thesis Diseño de un tablero de alertas tempranas para la evaluación inicial de clientes en el Banco Edwards: Una herramienta de apoyo para la optimización de tiempo del ejecutivo
No Thumbnail Available
Date
2026-10-13
Authors
Journal Title
Journal ISSN
Volume Title
Program
Ingeniería Comercial
Departament
Campus
Campus Casa Central Valparaíso
Abstract
La presente memoria tiene como objetivo diseñar un tablero de alertas tempranas para apoyar la evaluación inicial de clientes en el Banco Edwards, específicamente en la solicitud de créditos de consumo y créditos hipotecarios. El estudio surge en base a la práctica realizada en el Banco y la necesidad de los ejecutivos de optimizar el tiempo que utilizan al revisar antecedentes financieros, laborales y de comportamiento histórico antes de avanzar en el proceso de evaluación. El problema identificado no se relaciona con la falta de información, sino con la dificultad de reunir, ordenar e interpretar rápidamente los documentos relevantes de cada cliente. Por esta razón, se propone el desarrollo de un prototipo web de ingreso rápido, capaz de calcular indicadores básicos de riesgo y entregar una clasificación preliminar en tres niveles distintos: riesgo bajo, riesgo medio y alto riesgo. Para el desarrollo de la página web se revisaron antecedentes teóricos, normativos y financieros asociados al riesgo crediticio, la morosidad, evaluación de clientes y el uso de un tablero de control. Posteriormente, se definieron variables relevantes para la evaluación, tales como el ingreso líquido, carga financiera, cuota estimada, morosidad vigente, días de atraso, endeudamiento, antigüedad laboral, variabilidad de ingresos, personas a cargo y antecedentes educacionales o laborales. Estas variables fueron incorporadas en un sistema de puntaje ponderado, el cual permite calcular el peso de cada variable y entregar una alerta temprana sobre el perfil del cliente. El prototipo fue validado mediante casos simulados y una prueba piloto con 11 casos reales anonimizados proporcionados por el Banco Edwards, de las cuales 8 correspondieron a créditos de consumo y los 3 restantes de créditos hipotecarios. Debido a las políticas de confidencialidad del banco, no se usaron nombres, apellidos, RUT, direcciones ni antecedentes que permitieran identificar a los clientes evaluados. Los resultados obtenidos muestran que (...).
This thesis aims to design an early warning dashboard to support the initial evaluation of clients at Banco Edwards, specifically for consumer loan and mortgage loan applications. The study arises from the professional internship carried out at the Bank and from the need to optimize the time used by executives when reviewing financial, employment and historical behavior records before moving forward in the credit evaluation process. The problem identified is not related to the lack of information, but rather to the difficulty of gathering, organizing and quickly interpreting the relevant documents of each client. For this reason, the development of a fast-entry web prototype is proposed, capable of calculating basic risk indicators and delivering a preliminary classification into three different levels: low risk, medium risk and high risk. For the development of the website, theoretical, regulatory and financial background information related to credit risk, delinquency, client evaluation and the use of control dashboards was reviewed. Subsequently, relevant variables for the evaluation were defined, such as net income, financial burden, estimated installment, current delinquency, days past due, indebtedness, job seniority, income variability, dependents and educational or employment background. These variables were incorporated into a weighted scoring system, which makes it possible to calculate the weight of each variable and provide an early warning regarding the client’s risk profile. The prototype was validated through simulated cases and a pilot test with 11 anonymized real cases provided by Banco Edwards, of which 8 corresponded to consumer loans and the remaining 3 to mortgage loans. Due to the Bank’s confidentiality policies, names, surnames, national identification numbers, addresses or any other background information that could identify the evaluated clients were not used. The results obtained show that (...).
This thesis aims to design an early warning dashboard to support the initial evaluation of clients at Banco Edwards, specifically for consumer loan and mortgage loan applications. The study arises from the professional internship carried out at the Bank and from the need to optimize the time used by executives when reviewing financial, employment and historical behavior records before moving forward in the credit evaluation process. The problem identified is not related to the lack of information, but rather to the difficulty of gathering, organizing and quickly interpreting the relevant documents of each client. For this reason, the development of a fast-entry web prototype is proposed, capable of calculating basic risk indicators and delivering a preliminary classification into three different levels: low risk, medium risk and high risk. For the development of the website, theoretical, regulatory and financial background information related to credit risk, delinquency, client evaluation and the use of control dashboards was reviewed. Subsequently, relevant variables for the evaluation were defined, such as net income, financial burden, estimated installment, current delinquency, days past due, indebtedness, job seniority, income variability, dependents and educational or employment background. These variables were incorporated into a weighted scoring system, which makes it possible to calculate the weight of each variable and provide an early warning regarding the client’s risk profile. The prototype was validated through simulated cases and a pilot test with 11 anonymized real cases provided by Banco Edwards, of which 8 corresponded to consumer loans and the remaining 3 to mortgage loans. Due to the Bank’s confidentiality policies, names, surnames, national identification numbers, addresses or any other background information that could identify the evaluated clients were not used. The results obtained show that (...).
Description
Keywords
Página web, Riesgo crediticio, Modelos predictivos, Clasificación crediticia
