Data Mining para modelo predictivo de ventas y servicios de mantenimiento en un concesionario automotriz ligero

• Autores: Becerra Rojas, Jocelyn Pamela, Villarreal Roca, Enrique Martin
• Formato: tesis de grado
• Fecha de Publicación: 2021
• Institución: Universidad de Lima
• Repositorio: ULIMA-Institucional
• Lenguaje: español
• OAI Identifier: oai:repositorio.ulima.edu.pe:20.500.12724/15395
• Enlace del recurso: https://hdl.handle.net/20.500.12724/15395
• Nivel de acceso: acceso abierto
• Materia: Pronóstico de ventas; Minería de datos; Concesionarios de automóviles; Sales forecast; Data mining; Automobile dealers; https://purl.org/pe-repo/ocde/ford#2.02.04

Lately the level of competition between companies in the light automotive industry is reaching a very high level, due to the various strategies developed by many competitors. Our study seeks to strengthen the evaluation of forecasts to improve the organization's capability to anticipate future events in important business processes, such as sales and maintenance services. To achieve this objective, investigations related to Data Mining techniques were consulted, in order to perform an information analysis with a predictive approach. Our research involves designing different models applying methods such as regressions, neural networks and decision trees, to a historical database of an automotive organization, previously selecting data using techniques such as the correlation matrix and PCA (Principal Component Analysis). Finally, an evaluation is carried out on the results obtained after comparing the proposed models, where we find out that for sales forecasts, the neural network model implemented with PCA obtains better results; whereas, for maintenance services forecasts, the predominant model is the one implemented with Random Forest.