A machine learning pipeline for earthquake data analysis and prediction using graph-based deep learning. This project retrieves earthquake event data from the USGS Earthquake API, constructs a spatial graph of earthquake occurrences, and prepares the data for predictive modeling using PyTorch Geometric.

• 🔗 Fetch real-time earthquake data (GeoJSON) from USGS
• 📊 Convert raw data into a structured DataFrame
• 🌐 Build a K-Nearest Neighbors (KNN) Graph to connect earthquake events based on spatial proximity
• ⚡ Normalize features (latitude, longitude, depth, timestamp, magnitude) for ML training
• 🧠 Prepare graph datasets for use with Graph Neural Networks (GNNs) via PyTorch Geometric
• 📈 Scalable pipeline for earthquake prediction research

• Python 3
• Pandas / Fireducks (data handling)
• NumPy
• Requests (API calls)
• Matplotlib (visualization)
• Scikit-learn (scaling + KNN graph)
• NetworkX (graph creation)
• PyTorch + PyTorch Geometric (GNNs)

1. Install Dependencies

   pip install pandas numpy requests matplotlib networkx torch torch-geometric scikit-learn fireducks

2. Fetch Earthquake Data (USGS API)

   url = "https://earthquake.usgs.gov/fdsnws/event/1/query.geojson?...params..."
   response = requests.get(url)
   earthquake_data = response.json()

3. Convert Data → DataFrame

   • Extract longitude, latitude, depth, magnitude, and time
   • Convert timestamps to UNIX time

4. Graph Construction

   • Build KNN graph (K=5)
   • Nodes = earthquake events
   • Edges = nearest neighbor connections

5. Graph Neural Network Input

   • Normalize features with StandardScaler / MinMaxScaler
   • Create torch_geometric.data.Data object

   from torch_geometric.data import Data
   data = Data(x=x, edge_index=edge_index, y=y)

• Nodes (earthquakes): ~11,883

• Edges (spatial connections): ~38,483

• Data object ready for GNN models:

  Data(x=[11883, 4], edge_index=[2, 38483], y=[11883])
  

• Train GNN models to predict earthquake magnitude or occurrence likelihood
• Add temporal modeling for time-series predictions
• Visualize earthquake networks interactively

• USGS Earthquake API