Urban Mixer Maps
A dual-perspective map app connecting residents and tourists to explore urban diversity.
What I did
Machine Learning, User research,
Back-end and Front-end Design
Collaboration
Independent Project
Timeline
2 months, 2024 spring
Achievement
Research presented in 2024 ASCAAD Conference and published in its proceedings
Context
Cities naturally generate diversity, which is both economically efficient and culturally enriching. However, modern consumerism offers a facade of urban diversity, characterized by eye-catching appearances that are ultimately superficial,
chaotic, and meaningless. Orlando, Florida exemplifies this condition.
Problem
In Orlando, the spread of entertainment-driven aesthetics has overtaken the city’s appearance, prioritizing tourism while stripping resources from residents. Prolonged exposure to this sensory overload dulls local perception and intensifies the divide between visitors and the people who live there.
A later study shows tourists and residents' opposite preference of Orlando city appearance
Painpoint
Tourists and residents navigate the same city through completely different sensory and informational channels, resulting in fragmented experiences and missed opportunities for meaningful overlap.
Solution
I designed a system that analyzes urban-scape diversity from both objective and subjective lens to understand how tourists and residents emotionally respond to the city. The system will generate place recommendations and create community forums that help break down the segmentation between the two user groups.
An overview of analytical research work flow
Outcome
A closer examination of user group sentiment and their experiential relationship with the urban environment.
A multi-layered urban map
A new map APP bridges residents and tourists urban experiences.
Research + Design Process
Objective Visualization 01 - Urban Morphology Diversity
I implemented image classification model A (93% accuracy) to classify 2,856 satellite images into six categories. The results reveal that urban topology and spatial development are closely synchronized: residential zones tend to cluster around natural landscapes such as lakes, while amusement and industrial areas emerge around major infrastructure.
The region is predominantly residential, with infrastructure scattered throughout. Civic morphologies cluster in the northeast around downtown Orlando, while amusement morphologies concentrate on the opposite side near major tourist zones, and some extends into northwestern residential neighborhoods, reflecting local entertainment facilities for nearby residents.
Package Design
Available as a to-go companion and a home station, both made from durable molded pulp. Whether on a desk or in a pocket, the Pop pack is there when Glow getters need a quick reset.
Objective Visualization 02 - Urban Architectural Diversity
Image classification Model B was used to map architectural distributions across Orlando. Although amusement areas appear to exhibit artificial perspectival diversity, the analysis shows they are essentially homogeneous.
ResNet-34 and ResNet-50 were used to classify the requested GSV images into 11 architectural styles and map their spatial distribution for urban diversity analysis. The resulting map visualizes Orlando’s architectural diversity–homogeneity patterns.
raw GSV were requested from POI through Google Map API
null GSV data cleaning processs
Objective Visualization - Comparison
The visual comparison revealed that the level of urban homogeneity is almost opposite when comparing the city's fabric from a plan view to the street-level architectural elevations.
This contrast indicates that while tourist attractions appear morphologically rich and distinct from other contexts in Orlando when viewed from maps, this richness and distinction are perceived as homogeneity when observed from street-level views.
Discovery 01
Urban diversity is multi-layered, even when evaluated quantitatively.
Ultimately, its interpretation still depends on people’s perceptual perspectives.
Business Development
With GloWell, Kleenex shifts from volume to value introducing high-margin, lifestyle-driven products that elevate the brand and strengthen profitability and margins across retail and e-commerce.
Subjective Visualization - Emotional Response and Sentiment Prediction
Among the 50 residents surveyed through questionnaire, 46.96% found the street views of civic areas to be the most pleasant, followed by residential areas at 14.1%, and amusement areas at 12.18%. In contrast, only 5.57% and 5.08% of residents considered tourist areas, such as hotels and themed commercial areas, to be pleasant.
Using the Max-Diff algorithm, we concluded that there is a 77.73% probability that a local Orlando resident would prefer the diverse urban civic atmosphere over tourist and commercial areas, with the probability of them favoring commercial areas being only 29.28%. Reversely, tourists tended to value amusement areas, such as theme parks, the most at 45.13%. However, besides the themed parks as attractions for most people traveling to Orlando, overall preference order of the tourists closely resembled that of the residents. This suggests that while tourists appreciate the attractions and themed areas more, both groups share similar preferences for the pleasantness of civic urban environments.
Conversely, when we asked both residents and tourists to score urban diversity based on plan views, there was a consensus regarding the perceived diversity across different areas. Amusement areas were unanimously considered to exhibit the highest diversity, with a probability of 45.56%. This was closely followed by commercial areas, which garnered a probability of 19.84% , and Civic with a probability of 19.13%. Residential areas were perceived to have a moderate level of diversity, scoring 10.48% in the probability assessment, and industrial areas at only 5.43%.
Discovery 02
While residents tend to view major tourist areas negatively,
outside these attraction zones they share similar interests with tourists
The Urban Mixer App
Overlaying different layers of urban diversity maps (morphological and architectural) and spatial distribution map to form a context for urban sentiment prediction. This APP can be used for both tourists and residents.
