Twitter Data for Traffic Estimation

Abstract: Traffic estimation is a crucial aspect of transportation planning to anticipate increasing mobility and provide transportation infrastructure. Traditional methods, such as household travel surveys and traffic counting, often require significant time and financial resources. Therefore, this study explores the use of geolocation data from Twitter as an alternative for projecting trip generation in East Bandung, a rapidly developing area with new facilities such as the Gelora Bandung Lautan Api Stadium and the Summarecon Bandung commercial area. Data were collected using the Twitter API over one year (November 2023–November 2024) within a 700-meter radius and analyzed through spatial mapping and the development of an origin-destination (OD) matrix. Validation was conducted using the Mean Absolute Percentage Error (MAPE) and the Spatially Weighted Structural Similarity Index (SW-SSIM). The results identified Gedebage, Arcamanik, and Rancasari as the areas with the highest trip generation concentrations. A MAPE of 18.5% and an SW-SSIM of 0.72 indicate that Twitter data is sufficiently representative for modeling mobility patterns. Despite limitations such as population bias, Twitter data offers cost and time efficiency, making it a potential alternative to support more adaptive and real-time data-driven transportation planning in urban areas like East Bandung.