Publicação

Forecasting sales and transactions of fast-food stores: a proof of concept

Detalhes bibliográficos
Resumo:	As time goes on, more and more clients look for solutions to their data-related problems. During a 9-month internship at the Portuguese consulting company Noesis, a request was presented by a customer that wished to improve the forecasting capabilities of their fast-food chain, on sales and transactions, for four different distribution channels, and globally. Following a data analytics approach, hundreds of time series were examined, external variables were added, and two algorithms were used - ARIMA and Facebook’s Prophet. Both models were evaluated, and as each of them performed better in different segments, a hybrid system was implemented, successfully completing the task at hand. Based on the results, future improvements and recommendations were also identified.
Autores principais:	Mousinho, Cristina Isabel Palma
Assunto:	Machine Learning Forecasting demand Time series ARIMA Facebook Prophet
Ano:	2022
País:	Portugal
Tipo de documento:	dissertação de mestrado
Tipo de acesso:	acesso aberto
Instituição associada:	Universidade Nova de Lisboa
Idioma:	inglês
Origem:	Repositório Institucional da UNL

Descrição
Resumo:	As time goes on, more and more clients look for solutions to their data-related problems. During a 9-month internship at the Portuguese consulting company Noesis, a request was presented by a customer that wished to improve the forecasting capabilities of their fast-food chain, on sales and transactions, for four different distribution channels, and globally. Following a data analytics approach, hundreds of time series were examined, external variables were added, and two algorithms were used - ARIMA and Facebook’s Prophet. Both models were evaluated, and as each of them performed better in different segments, a hybrid system was implemented, successfully completing the task at hand. Based on the results, future improvements and recommendations were also identified.