SARIMA-ELM Hybrid Model for Forecasting Tourist in Nepal

In this study a novel hybrid model has been developed to forecasting tourist arrivals. The main concept is to combine two different forecasting techniques such as SARIMA and Extreme Learning Machine models to produce a new SARIMA-ELM hybrid Model, so as to achieve accuracy in forecasting. Forecasting accuracy for SARIMA, Triple Exponential Smoothing (The Holt-Winter’s), Multi Layer Perceptron-Neural Networks (MLP-NN), Extreme Learning Machine (ELM) and SARIMA-ELM hybrid models are computed and compared using criteria like RMSE, MAE, and MAPE. Empirical analysis found that SARIMA-ELM hybrid model has highest forecasting accuracy Thus, SARIMA-ELM hybrid model is the most appropriate model for forecasting tourist arrivals.