Predicting Air Quality in Wildfire Zones

Project Overview

• Developing machine learning models to predict air quality using diverse data sources
• Created two supporting R packages to streamline research workflow
• Processing and analyzing massive environmental and traffic datasets

Components

1. Air Quality Prediction Model

   • Data cleansing and preparation
   • Exploratory data analysis
   • Geospatial operations
   • Dataset integration
   • Feature engineering
   • Processing 100M+ rows of environmental data
   • View Project

2. Custom R Packages

   • purpleAirAPI: Streamlined PurpleAir data collection
   • DataOverviewR: Automated EDA and reporting

Technical Stack

• Languages: R
• Key Libraries: sf, dplyr, ggplot2, randomForest
• Data Sources: PurpleAir sensors, OpenStreetMap, Uber Movement, Air Quality System (US EPA), Wildland Fire Perimeters (CalFire), Weather (Iowa Environmental Mesonet)