Inspiration
Cities face repeated problems such as flooding, traffic disruption, and infrastructure failures, but every city reacts differently. We were inspired by the idea that a city can have its own behavioral pattern, just like a unique DNA. Instead of only detecting problems, we wanted to understand how a city behaves and predict what could happen next.
What it does
CityDNA AI learns from historical and real-time city data to build a unique behavioral profile of a city. It predicts how the city may respond to events such as heavy rainfall, road closures, or infrastructure failures and provides early insights to help authorities take proactive action.
How we built it
We combine historical urban data, weather information, traffic patterns, infrastructure records, and citizen reports. Machine learning identifies recurring patterns, while prediction and visualization modules display risk areas, expected impacts, and recommendations through a simple dashboard.
Challenges we ran into
The biggest challenge was combining different types of urban data and finding meaningful relationships between them. Another challenge was ensuring that predictions remain reliable when real-world city data is incomplete or inconsistent.
Accomplishments that we’re proud of
We developed a concept that goes beyond traditional event prediction by creating a city-specific behavioral intelligence system. The platform can continuously learn from new events and improve its understanding of how the city responds.
What we learned
We learned that smart cities are not just about collecting data. The real value comes from understanding relationships between different systems and turning data into actionable decisions.
What’s next for CityDNA
We plan to integrate more real-time data sources, improve prediction accuracy, add more urban scenarios such as cyclones and heatwaves, and eventually expand the platform to learn the unique behavioral signatures of different cities.
Built With
- fastapi
- gis
- graph-analysis
- machine-learning
- postgis
- postgresql
- python
- react.js
- time-series-forecasting