| Autor: |
Lavrentios Vasiliadis, Dimitrios I. Vortelinos, Konstantinos Gkillas, Christos Floros |
| Rok vydání: |
2020 |
| Předmět: |
|
| Zdroj: |
Operational Research in Agriculture and Tourism ISBN: 9783030387655 |
| DOI: |
10.1007/978-3-030-38766-2_6 |
| Popis: |
We study the performance of the k nearest neighbor (kNN) forecasts in the context of European tourism demand. The forecasting performance of neural networks is examined across different parameterizations of the kNN model. The selection of the most appropriate kNN parametrization can produce more accurate forecasts. Tourism demand is forecast monthly for 20 European countries. Tourism demand is measured via seven variables for the reason of consistency in results. kNNs better forecast tourism demand in shorter horizons; in specific, 1 month ahead. The parametrization of the kNN model affects forecasting performance. More sophisticated parameterizations perform better than either an ARIMA model or a naive kNN parametrization. The inclusion of international stock indices significantly increases forecasting accuracy. The more explanatory variables employed, the higher forecasting accuracy is retrieved. However, there is not a specific group of stock markets affecting more the kNN model’s forecasting accuracy. The forecasting accuracy of kNNs differs between three (Western, Eastern and Southern) European regions. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
