QuakeGuard

QuakeGuard Home Page

Inspiration

Our inspiration for this project rose from the devastating earthquakes that struck Turkey recently. As environmental enthusiasts, we knew we wanted to help. Our goal was to make something new and exciting that combined elements we were familiar with such as Node.js, React.js, and TailwindCSS, as well as elements we wanted to learn more about such as Machine Learning. And so began our journey to create an extraordinary hack that not only addresses a current problem but also paves the way for a brighter future.

What it does

QuakeGuard relies on Earthquake datasets, pandas analysis, and a ML model to accurately tell the user the chances that an earthquake will occur in their area and its expected strength

So... How are we DIFFERENT???

AI and Machine Learning Powered Uses AI and Machine Learning to predict earthquakes with higher accuracy, enabling timely and informed decisions by authorities and communities!
Global Community & Support Unlike traditional models, QuakeGuard involves an active amount of users to allocate resources to high-risk areas, helping the ones who most need it!
Holistic Analysis Our analysis considers longitude and latitude, depths of earthquakes, dates, and Richter ratings from a diverse dataset providing more accurate outputs and comprehensive summary results!

How we built it

It was built on VS code, running on a NextJS/Flask stack. During the hackathon, we split up into the areas we were most comfortable too but also willing to understand more. One backend, one front end, one ML, and a final designer that later shifted to a front-end developer later on in the development cycle. We worked on an agile workflow responding to changing needs or market conditions like the workshops and fun events happening in Delta. The ML model was made using an external database and used Flask to work with the backend code in the function. We finally, deployed it on Google Cloud.

Challenges we ran into

Our biggest and primary challenge was making the ML model, both implementing it and integrating it with our program. Finding the ML dataset was tedious and we found a hard time using it to find accurate results as even a minute latitude or longitude change can make a major impact on the risk factor of an earthquake. As well, once we were satisfied with the accuracy, we found it hard to connect this external set of data to our backend program. After a while, using Flask, we were able to make do but it was too big of a file to perform push and pull in vs code making it harder to test.

Accomplishments that we're proud of

Implementing all the items we initially brainstormed to work well in the end.
Using TailwindCSS, and React to make a dynamic website that is user-friendly and aesthetic.
Getting the ML model working and running, outputting clear and useful data, which can help authorities allocate resources. I learned that letting each person work on their area of interest is good as it boosts their morale.

What we learned

All members of our team learned new languages/libraries and frameworks, which is a good experience because a) it's fun, b) skill booster, and c) a social avenue with people. One surprising thing I found out was that the most complex solution might not be the only/best solution as simple concepts can work wonders.

What's next for QuakeGuard

Connect with backend servers such as MongoDB or MySQL to save password data so users can save their payments multiple times or give annual payments throughout the year.
Use features like Google API and Stripe to collect online transactions
Be more accurate with ML testing to make sure small latitude/longitudinal changes do not change it too much.
Using the date data given, predict around what time an earthquake would be common to happen in an area.