Integrating Google Search Trends with Advanced Predictive Models for Forecasting German Tourist Arrivals to Turkiye: A Decade-Long Analysis

Abstract

Over recent decades, tourism has played an increasingly vital role in the socioeconomic development of various regions, including Turkiye. This study focuses on forecasting the number of German tourists entering Turkiye every month over the past decade. The dataset, obtained from the Ministry of Culture and Tourism, includes the monthly number of German passport holders entering Turkiye, with data spanning ten years from January 2014 to November 2024 that incorporates the COVID-19 pandemic. Additionally, correlations between these numbers and the popularity of selected Google search keywords from Germany are analyzed. Predictive models are developed by utilizing the search popularity of these keywords alongside monthly tourist entry data. The models employed include Artificial Neural Networks (ANN), Long Short-Term Memory (LSTM), Recurrent Neural Networks (RNN), Seasonal Autoregressive Integrated Moving Average Exogenous (SARIMAX), Support Vector Regression (SVR), and PROPHET. The inclusion of pandemic data offers a comprehensive perspective on the impact of global disruptions on travel trends. Initial results highlight the effectiveness of these models in capturing complex patterns and seasonality in the data, paving the way for more accurate and dynamic forecasting methods. This study contributes to the development of a practical tool for tourism analytics and emphasizes the potential of integrating search engine data into predictive models for enhanced decision-making in the tourism sector.