Portfolio-grade renewable energy platform
Wind energy forecasting from distributed Spark pipelines to live NOAA analysis.
This project combines large-scale NOAA weather processing, PySpark ETL, turbine-inspired wind modeling, ML forecasting, Airflow orchestration, benchmarking, preserved website artifacts, and a deployable FastAPI live analysis backend.
Historical Window
1995–2025
Verified Live Stations
1,981
Forecast Evaluation Rows
535,961
Live Wind Outlook
Explore live NOAA observations, turbine-inspired power-curve estimates, and deployable backend wind outlook analysis.
Pipeline Architecture
Review the Spark ETL pipeline, Airflow orchestration, feature engineering, ML workflow, and artifact preservation design.
Historical Results
Analyze long-run wind potential trends, state summaries, regional outputs, and historical Spark analytics.
Forecasting Model
Inspect model metrics, holdout forecast evaluation, feature importance, and forecasting diagnostics.
Benchmarking
Compare Spark and DuckDB analytical execution performance across benchmark workloads.
End-to-end forecasting workflow
ingestion → cleaning → feature engineering → forecasting → preserved artifacts → live analysis