Interactive Visualization of Tourism Sentiment Analysis as a Strategic Decision Support Tool for Local Governments

Abstract

The tourism sector of Sumedang Regency has experienced significant growth following the increased accessibility of the region after the operation of the Cisumdawu Toll Road. This growth has been followed by a large number of tourist opinions spread across digital media, such as Google Maps. This data has the potential to be a valuable source of insight for the local government if processed systematically. This research aimed to conduct sentiment analysis on tourist reviews and design an interactive dashboard as a strategic decision-support tool. Data was collected from 23 tourist destinations in Sumedang over the last year, resulting in 695 reviews in Bahasa Indonesia. The pre-processing steps included case folding, cleaning, tokenizing, stopword removal, and stemming. The Support Vector Machine (SVM) algorithm was used to classify review polarity into positive and negative categories. The model testing results showed an overall accuracy of 92.8%. However, further analysis revealed challenges due to imbalanced data: the model performed excellently in recognizing positive sentiment (99% Recall) but had lower performance in detecting negative sentiment (65% Recall). The classification results were then visualized using Microsoft Power BI, incorporating geospatial map components, sentiment comparison bar charts, and sentiment proportion pie charts. The resulting dashboard can dynamically display the distribution of tourist sentiment per destination and tourism category, making it easier for local government to monitor public perception and set data-driven tourism development priorities.