Forecasting performance of cruise passengers : the Spanish ports case

This contribution examines the passenger forecasting performance of the SARIMA method applied to cruise activities in the main Spanish ports. In this port system, the cruise activity market is characterized by different seasonal patterns (i.e., once- or twice-yearly peaks, which means unimodal or bi...

Descripción completa

Detalles Bibliográficos
Autores: Grifoll Colls, Manel|||0000-0003-4260-6732, Sánchez Espigares, Josep Anton|||0000-0001-8195-1913, Feng, Hongxiang
Tipo de recurso: artículo
Fecha de publicación:2021
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/343809
Acceso en línea:https://hdl.handle.net/2117/343809
https://dx.doi.org/10.1002/jtr.2433
Access Level:acceso abierto
Palabra clave:Transportation -- Passenger traffic
Passenger ships
Bimodal
cruise passengers
forecasting
seasonality
spanish ports
time series
Transport de viatgers
Vaixells de passatge
Àrees temàtiques de la UPC::Nàutica::Navegació marítima::Transport marítim
Àrees temàtiques de la UPC::Economia i organització d'empreses::Direcció d'operacions::Modelització de transports i logística
Descripción
Sumario:This contribution examines the passenger forecasting performance of the SARIMA method applied to cruise activities in the main Spanish ports. In this port system, the cruise activity market is characterized by different seasonal patterns (i.e., once- or twice-yearly peaks, which means unimodal or bimodal behavior) due to repositioning strategy. The outcome of standard indicators for accuracy testing reveals inconsistent prediction performances among ports. These inconsistencies are analyzed using an index of bimodality and seasonal variability. The forecasted values for a high-level of bimodality and seasonal variability show worse prediction performances than unimodal patterns and low seasonal variability. Ports with less passenger activity entail larger predictions errors. Exponential and linear models were adjusted between the error metrics and the mentioned indexes.