Hyperparameter fine tuning for a time series forecasting model

• Autor(a) principal: Magalhães, Manuel Maria Da Cunha
• Data de Publicação: 2021
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10362/145522
• topic: Business analytics; Business and data analytics; Grid search; Hyperparameter fine tuning; Random search; Domínio/Área Científica::Ciências Sociais::Economia e Gestão

This project was conducted in the context of the Project-Based Learning program. The purpose of the program is to provide an experience in a real-life business and data analytics project. During the last 18 months a work collaboration have been carried out between four NOVA SBE Business Analytics master students and Brisa. The main objective of the project was to produce new traffic forecasting models in Python. The individual work carried out by the author of this study, was focused on the hyperparameter fine tuning procedure for the forecasting models. The research for different methodologies resulted in the experimentation of grid search and random search frameworks. As expected, grid search achieved better results but it is a process that requires more computational power and time.