SF Flood Prediction

Our project predicts the flood vulnerability index based on user adjustable features like elevation, sea level, poverty, and education. It is based on the lightGBM model.

Comment

Inspiration

We were interested in examining how environmental conditions affect health outcomes in the SF community. Environmental factors like flooding and precipitation can be a significant social determinant of health, so understanding flood vulnerability can help us identify how poor environmental outcomes disproportionately affect vulnerable populations.

What it does

This predicts the Flood Vulnerability Index using factors like elevation, sea level, disability, mental health, asthma, and more. Users can adjust the sliders to view how these factors influence the flood vulnerability index.

How we built it

To start, we used Pykaret to compare different predictive models. We found that the lightGBM model predicted at the highest accuracy with an R2 of over .96. So we saved the lightGBM model and loaded it into Streamlit. Then we added all the sidebars which included the user adjustable features. The scales were based on ranges shared in the raw data. The output is a function that takes the user's input feature and runs the value through the lightGBM model to compute the flood index value.

Challenges we ran into

A major challenge was the time constraint and competing priorities.

Accomplishments that we're proud of

Using a machine learning model from the data

What we learned

We learned how to use Streamlit and machine learning to analyze and share data

What's next for SF Flood Prediction

Finding a permanent url for this project

Built With

Share this project:

Updates