Tourism forecasting using hybrid modified empirical mode decomposition and neural network
Due to the dynamically increasing importance of the tourism industry worldwide, new approaches for tourism demand forecasting are constantly being explored especially in this Big Data era. Hence, the challenge lies in predicting accurate and timely forecast using tourism arrival data to assist gover...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
International Center for Scientific Research and Studies
2017
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/66467/1/RuhaidahSamsudin12017_TourismForecastingusingHybridModified.pdf http://eprints.utm.my/id/eprint/66467/ http://home.ijasca.com/data/documents/Vol_9_1_ID-14_Pg14-31.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Due to the dynamically increasing importance of the tourism industry worldwide, new approaches for tourism demand forecasting are constantly being explored especially in this Big Data era. Hence, the challenge lies in predicting accurate and timely forecast using tourism arrival data to assist governments and policy makers to cater for upcoming tourists. In this study, a modified Empirical Mode Decomposition (EMD) and Artificial Neural Network (ANN) model is proposed. This new approach utilized intrinsic mode functions (IMF) produced via EMD by reconstructing some IMFs through trial and error method, which is referred to in this research as decomposition. The decomposition and the remaining IMF components are then predicted respectively using ANN model. Lastly, the forecasted results of each component are aggregated to create an ensemble forecast for the tourism time series. The data applied in this experiment are monthly tourist arrivals from Singapore and Indonesia from the year 2000 to 2013 whereby the evaluations of the model’s performance are done using two wellknown measures; RMSE and MAPE. Based on the empirical results, the proposed model outperformed both the individual ANN and EMD-ANN models. |
|---|