Modelo de aprendizaje automatizado del proceso de venta de productos financieros en un Call Center

• Autores: Gutierrez Salas, Jorge Joao, Vigo Liñan, Vanessa Stephany
• Formato: tesis de grado
• Fecha de Publicación: 2021
• Institución: Universidad de Lima
• Repositorio: ULIMA-Institucional
• Lenguaje: español
• OAI Identifier: oai:repositorio.ulima.edu.pe:20.500.12724/14344
• Enlace del recurso: https://hdl.handle.net/20.500.12724/14344
• Nivel de acceso: acceso abierto
• Materia: Aprendizaje automático; Inteligencia artificial; Productos financieros; Ventas; https://purl.org/pe-repo/ocde/ford#2.02.04

This project focused on the design, construction and implementation of a service that predicted the behavior of a potential client, in order to finalize the sale of a financial product in advance, which is based on machine learning. The prototype has been tested with a database of clients of a financial institution, which were offered a financial product such as, for example, a freely available loan, credit cards, loans to pymes, mortgage loans, vehicle loans, etc. and obtaining as a result a final sale or a rejection of the offer. With this information, and through different predictive algorithms, an adequate model was built to predict sales of financial products. The beneficiaries of the implemented solution will be companies that provide outsourcing services (BPO) to financial entities. These companies obtain profits from sales commission using human and technological resources to achieve sales. Under this scheme, the predictive model implemented made it possible to make a service available which, when invoked, allows to increase the probability of sale and in turn managed to optimize the operation at the human resource level, reducing the number of sales executives, and increasing productivity. of the back office area of the outsourcing service, avoiding downtime typical of an operation dependent on the completed sales. The prediction module developed was presented in a web application that allowed the input data to be entered (historical sales record) and as a result the predictions based on the machine learning models that obtained the best results were shown. The prediction and its evolution over time were presented in an interactive dashboard showing the monthly sales results, sales campaigns, number of leads, predictive models, sales effectiveness, sales per month and total sales.