Neuro-Fuzzy Forecasting of Tourist Arrivals

This study develops a model to forecast inbound tourism to Japan, using a combination of artificial neural networks and fuzzy logic and compares the performance of this forecasting model with forecasts from other quantitative forecasting methods namely, the multi-layer perceptron neural network model, the error correction model, the basic structural model, the autoregressive integrated moving average model and the naive model. Japan was chosen as the country of study mainly due to the availability of reliable tourism data, and also because it is a popular travel destination for both business and pleasure. Visitor arrivals from the 10 most popular tourist source countries to Japan, and total arrivals from all countries were used to incorporate a fairly wide variety of data patterns in the testing process. This research has established that neuro-fuzzy models can be used effectively in tourism forecasting, having made adequate comparisons with other time series and econometric models using real data. This research takes tourism forecasting a major leap forward to an entirely new approach in time series pedagogy. As previous tourism studies have not used hybrid combinations of neural and fuzzy logic in tourism forecasting this research has only touched the surface of a field that has immense potential not only in tourism forecasting but also in financial time series analysis, market research and business analysis.

Name dc:type.qualificationname

phd

Level dc:type.qualificationlevel

doctoral

Grantor dc:publisher.institution

Victoria University

Year dc:date.issued

2005

Author dc:creator

• Fernando, Hubert Preman

• School of Economics and Finance
• 350000 Commerce, Management, Tourism and Services

Language dc:language

en