Diseño de un modelo de aprendizaje automático para gestion de la flota vehicular de una Empresa Pública.

Abstract

This study presents the design of a machine learning model for managing the vehicle fleet of a Public Enterprise, focused on classifying drivers according to their level of risk during driving. To achieve this, driver characteristics, their behavior and the use of vehicles are analized to generate optimization recommendations for the fleet. Web scraping plays an important role in the development of this work, for obtaining data from drivers, vehicle fleet and its usage, which in turn facilitates the traditionally manual process of vehicle registration validity control performed by the Enterprise. In Chapter I, the generalities of this project are stated, its objectives, justification, and importance, and a literature review on web scraping, machine learning models and Information Systems (IS) are presented. In Chapter II, the methodology to be used is described, a descriptive analysis of the fleet, driver characteristics and behavior, and vehicle usage by departments and management to generate fleet optimization recommendations. It also involves the development of driver classification models and finally, a dashboard is developed to share the results of the fleet analysis, driver analysis, generated model and a map for visualizing vehicle usage. In Chapter III, the results obtained are presented and analyzed. Finally, in Chapter IV, the conclusions derived from the project development are detailed.