Answers are grounded in this paper and cite their sources.
Develops floating population estimation model, validating effectiveness in urban areas, suggesting public data is reliable.
Despite the growing demand for data-driven and systematic urban planning, even basic population data is not being provided in a timely manner. To address this issue, this study developed a floating population estimation model using public data and validated its applicability. As a pilot case, the model was applied to estimate the average daily floating population in the central and southern districts of Gimhae City, South Korea. The core dataset was intelligent CCTV people-counting data collected and managed by public institutions. Independent variables included a range of urban spatial factors, such as population, land use, building use, and other factors. To construct the optimal estimation model, machine learning regression algorithms were applied, with RMSE used as the primary evaluation metric. The results identified Gaussian Process Regression as the optimal model for the central district and Linear Support Vector Machine for the southern district, both trained and validated using an 80:20 data split. The comparison between the estimated floating population data and the telecom-based floating population data yielded MAE of 0.2825 and 0.2854 for the central and southern districts, respectively, demonstrating a relatively high level of predictive accuracy. These findings confirm the potential of public data-based floating population estimates as a viable alternative to the telecom-based floating population data. In the future, by incorporating dynamic variables such as traffic volume, weather conditions, and event information, the model could be extended to real-time population estimation and effectively support urban policy-making and decision-making processes.
No takes yet. Share an insight, caveat, or question.
Kim et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: