Final project made for the AI In Urban Planning And Design 2 class at Seoul National University.
This repository includes the notebooks used to:
- Make clusters of cities from 50 american cities depending on their urban form
- Train ML models to determine the air quality (using concentration of different chemicals in the air) from daily report including date, meteorological informartion, traffic and nearby polluting power pants.
- Evaluate the importance of the different features depending for each cluster, to highlight the differences between them.
This repository also includes the slides for the project proposal and the final report giving an overview of the results. It does not include the data used for the analysis. The daily air quality report can be found on kaggle (https://www.kaggle.com/datasets/mayukh18/deap-deciphering-environmental-air-pollution).
While the study is an interesting start, the lack of time limited the insight it could give. In particular, getting more information about the cities' urban form to have more interesting clusters is probably the most important step.