RF: Earthquake Predictor

• 

  this map ui shows prominent wildfire locations

Inspiration

Recently, the city of Dublin, California experienced an unexpected and unsettling event: two back-to-back earthquakes that caught residents completely off guard. These seismic events, although not catastrophic, served as a wake-up call, highlighting the unpredictability of earthquakes and the lack of accessible, real-time tools to help people understand or anticipate them.

What it does

This project is designed to monitor, analyze, and visualize global earthquake activity using real-time data from the US Geological Survey (USGS). By automatically retrieving up-to-date information on earthquakes occurring around the world for the past 30 days, the system enables continuous tracking of seismic events. The goal is to transform raw earthquake data into accessible, meaningful information that can support public awareness, scientific research, and potential early warning or risk assessment efforts.

How we built it

We used Python to pull real-time earthquake data from the USGS, cleaned it, and visualized it using Seaborn and Folium. Then we trained a Random Forest model to classify whether an earthquake is significant, using SMOTE to handle class imbalance. We also built an interactive map and evaluated model performance with metrics like accuracy and F1 score.

Challenges we ran into

The main issues were class imbalance, occasional missing data from the live feed, and slow map performance with lots of points. We used SMOTE for balancing, added checks in preprocessing, and solved the map issue with clustering. Choosing between classification and regression was also a key early decision.

Accomplishments that we're proud of

We are proud of creating a real-time data pipeline and using forest random to integrate machine learning into our application. We are also proud of creating an interactive global map.

What we learned

Throughout the development of our Earthquake Detector, we gained valuable experience in several key areas such as machine learning implementation, working with real time data and data processing

What's next for RF: Earthquake Predictor

For RF: Earthquake Predictor, we could improve it by making the application predicting earthquakes with a magnitude greater than 6.0, using a different learning model to potentially increase our F1 Score, improving our data visualization by adding filters to the folium map and making the transition from graph to map smoother

Built With

• css
• data-processing
• development-deployment
• html
• javascript
• machine-learning
• python
• visual-studio
• visualization
• web-data-access