Smart City: Shopping Area Diversity

Observation

Vendors and shopkeepers in Tainan's old shopping areas often placed equipment and tables on sidewalks, making them too narrow to walk through. This created both accessibility and aesthetic challenges. Rather than pursuing temporary aesthetic fixes, this project sought to bring a science-informed approach to design by identifying prototypical street patterns that could guide sustainable improvements.

Research Range

The study covered four districts in Tainan: Anping, West Central, North, and East.

Data Analysis: Stores' Data

Over 7,000 data points were collected from Google Maps across seven store categories: shopping malls, restaurants, museums, historical sites, convenience stores, cafes, and bars. Six shopping areas were selected for deeper analysis.

Computer Vision Analysis

Using YoloV3, street-level objects were detected across the shopping areas, including stalls, vendor umbrellas, awnings, and shop signs.

Correlation Analysis

The study examined correlations between store types and street elements. Awnings showed a positive correlation with shopping malls and cafes, while restaurants displayed a negative correlation with awnings — potentially due to differences in visitor dwell time.

Transportation Data

Walking range analysis assumed a 300-meter accessible radius, accounting for Tainan's hot and humid conditions. Three modes of transit access were mapped: bus stops, T-bike stations, and parking lots.

Conclusion

Six shopping area prototypes emerged, each categorized by its level of diversity. The analysis revealed distinct patterns: traditional markets showed higher mobility and narrower spaces, tourism-focused areas had more complex store mixes, and attraction-based areas centered on historical sites.

Future Urban Plan

Three pilot interventions applied these findings to specific shopping areas, demonstrating how quantitative urban analysis can inform place-specific design grounded in existing street fabrics rather than imposing standardized solutions.