Due to the recent recurring events of natural disasters such as the Manghut Typhoon and wildfires in california, we decided to contribute to the rescue missions with machine learning.
What it does
- By using chatbot as an interface, it collects critical informations from the users and assist them with suggestions.
- We use Kmean Clustering algorithm to find multiple optimized coordinates to send mass transportation vehicles and visualizes it on an admin dashboard.
- The users can also query for the recent weather to check if there is going to be an alerts.
How we built it
We developed a chat bot using reactjs and google cloud platform - natural language and used accuweather APIs to get weather data and alarms. We store the data on sqlite3 with a backend server running ruby on rails and nodejs. Machine learning model, Kmean Clustering algorithm, is built with sklearn in python.
Challenges we ran into
We had some issues integrating the various systems. Making the UI simple on the frontend with complex data analysis, processing and communicating with various platforms on the backend was very challenging.
Accomplishments that we're proud of
The complex structure we built for this project.
What we learned
- The experiences we gain from teamwork.
- The usage of Google Cloud Platform and Accuweather's API.
- Applications of machine learning models.
- The experience of building a production application which can help so many people.
What's next for Natural Disaster Bot (NDB)
- Improve our machine learning model and add more features for our relief suggestion tools.
- Make it more scalable and accessible.