Tourism Analysis and Prediction in the Context of Nepal

Abstract

Tourism is a vital pillar of Nepal’s economy, offering substantial contributions to employment, foreign exchange earnings, and regional development. This research presents a data-driven approach to analyze and forecast Tourist arrivals in Nepal using advanced statistical and machine learning models. Historical tourism data spanning from 1992 to 2017 was obtained from the Ministry of Culture, Tourism and Civil Aviation (MoCTCA), the World Travel & Tourism Council (WTTC), and other survey-based sources. The collected data underwent rigorous preprocessing, including cleaning, scaling, transformation, and decomposition, to ensure suitability for predictive modeling. Three forecasting models were implemented: Simple Linear Regression, Seasonal ARIMA (SARIMA), and Multi-layer Perceptron (MLP). The SARIMA model captured seasonal trends in monthly Tourist arrivals, while the MLP model integrated multivariate inputs—such as accessibility, accommodation, and healthcare infrastructure—for enhanced nonlinear forecasting. Performance was evaluated using metrics like RMSE and MSE. Among the models, the MLP achieved the highest predictive accuracy, effectively modeling the complex relationships and patterns in the data. These results offer valuable insights for policymakers and tourism stakeholders to optimize planning, marketing, and infrastructure development strategies based on robust forecasts.