In recent years, with the widespread use of location data, expectations are rising for urban analysis and tourism management that utilize human flow data.
This study analyzed the distribution of visitors by age, gender, and weekday/holiday using 250m mesh pedestrian flow data in the central area of Kobe City. First, Getis-ord Gi*We extracted hotspots and coldspots using statistical measures and quantified the spatial distribution of each attribute. Subsequently, we performed JMP cluster analysis using the Gi* statistic as a feature to categorize spatial pedestrian flow patterns.
As a result, distinctive clusters were identified for each demographic, including a weekend-type pattern mainly among young people, a middle-aged and older type with a high proportion of weekday business use, and an independent pattern mainly among the elderly.Furthermore, a comparison with classification using the Image Similarity Index (FSIM) showed that cluster analysis directly using the Gi* statistic yielded more interpretable results that reflected the spatial structure .
This study demonstrates that the characteristics of human flow data can be clearly visualized by performing cluster analysis using JMP, and is expected to be useful in tourism policies, urban planning, and accommodation location strategies.
