Pranav Kuchibhotla, Obaid Bin-Mahfoudh, Sebastian Buzenas, Gavin Coffer, Thailer Simmons
Illinois Institute of Technology, Spring 2025 – IPRO 497
Chicago has over 400,000 lead service lines—the most in any U.S. city. Lead exposure, especially in children and vulnerable populations, causes irreversible health effects. Our project aims to:
- Visualize geospatial lead contamination risk
- Analyze seasonal and temperature-driven trends
- Prioritize equitable service line replacement using public health data
🔗 Live demo: 82-below.vercel.app
- Geocoded 400k+ addresses from city inventory and matched with assessor data
- Calculated Cost of Lead (CoL) and Likelihood of Lead (LoL) scores at the census block group level
- Final Risk Score =
CoL × LoL
- Conducted statistical hotspot mapping across:
- % of youth population
- % uninsured
- Median income
- % suspected/public/private lead lines
- Combined to highlight tracts where equity and infrastructure risks intersect
- Used 10 years of lead sample test data
- Bootstrapped 90th percentile lead levels across seasons
- Found statistically significant increases in summer/fall, peaking in September–October
- Validated using permutation tests and Euclidean distance metrics
- Animated choropleths of rolling 30-day lead 90th percentile levels by ZIP
- ZIP-level seasonal risk maps (
Warm Lead90 / Cold Lead90) - Property age risk curves, residence-type heatmaps, and equity overlays
We created a public-facing website to help Chicago residents visualize nearby lead service lines and understand their local risk.
🔗 Live Site: CHECK YOUR ADDRESS
Key Features:
- 🔎 Search by Address — View nearby service line data for any Chicago location
- 🗺️ Interactive Map with markers for:
- 🔴 Lead / Galvanized Requiring Replacement
- 🟢 Copper / Not Lead
- ⚪ Unknown
- 📍 Details Panel with:
- Distance from the searched location
- Pipe material (Public/Private)
- Classification of each line
- 🎯 Adjustable search radius (1–20 nearest lines)
Example Result:
Search for 1050 N Harding Ave, Chicago, IL 60651 returns 5 nearby service lines, all classified as Lead:
| Address | Classification | Public Material | Private Material | Gooseneck/Pigtail | Distance (ft) |
|---|---|---|---|---|---|
| 1100 N HARDING AVE | Lead | L | U | L | 145 |
| 1058 N HARDING AVE | Lead | L | L | L | 177 |
| 1103 N HARDING AVE | Lead | L | U | L | 174 |
| 1056 N HARDING AVE | Lead | L | U | L | 189 |
| 1106 N HARDING AVE | Lead | L | U | L | 206 |
📍 Location Coordinates:
Latitude: 41.900553
Longitude: -87.725318
This result highlights how the tool empowers residents in high-risk zones to discover real risks based on public data.
Why it matters:
With over 400,000 lines in Chicago, this tool offers residents personalized transparency and encourages action—especially in high-risk zones.
📁 data/ # Raw datasets (FOIA, census, assessor, test results)
📁 scripts/ # Processing, scoring, cleaning (Python)
📁 notebooks/ # EDA, time series, statistical testing
📁 visualizations/ # Heatmaps, choropleths, rolling GIFs, static plots
📁 docs/ # Midterm, final reports, presentation assets
README.md # You are here
Due to GitHub file size limitations, our complete raw datasets, GIS files, high-resolution maps, posters, and presentation materials are hosted externally.
📁 Access full project assets via OneDrive:
🔗 IPRO 497 S25 – Data Visualization Folder (OneDrive)
Includes:
- FOIA files and lead test inventory
- Census & ADI demographic data
- Final presentation PDFs and posters
- Animated maps, ArcGIS files, ZIP-level risk tables
Use this folder if you'd like to reproduce the full analysis or access high-res visuals that aren’t stored in the GitHub repo.
We employed a combination of spatial analysis, statistical testing, and temporal modeling techniques to build a robust, data-driven understanding of lead contamination in Chicago:
- 📍 Kernel Density Estimation (KDE): Used to generate both static and interactive maps showing spatial clusters of lead service lines, highlighting high-density areas across the city.
- 🌡️ Bootstrap Confidence Intervals: Seasonal lead sample data was bootstrapped 10,000 times to produce reliable, non-parametric confidence intervals for both the mean and 90th percentile lead levels.
- 🎲 Permutation Testing: Conducted 1,000 permutations to statistically validate the alignment of temperature trends and lead level peaks, confirming a ~2-month lag effect in seasonal risk.
- 🧪 Risk Scoring Framework: Developed a composite score (
Risk = CoL × LoL) using social vulnerability indicators (ADI, children under 5) and lead likelihood estimates from infrastructure data. - 🗺️ Getis-Ord Gi Hotspot Analysis*: Applied spatial statistics to identify tracts with significantly higher vulnerability and lead presence, combining multiple demographic and infrastructure layers.
- 🎞️ Animated Time Series Visualizations: Rolling 30-day 90th percentile maps were animated to reveal lead level evolution over time, emphasizing seasonality and data gaps in specific ZIP codes.
These methods allowed us to triangulate where, when, and why lead exposure is most dangerous — forming the basis for our prioritization and public outreach strategies.
- 2172: Estimated full replacement timeline with current budget ($16,000/line, $53M/year)
- Summer = Risk: Lead90 peaks lag temperature by ~2 months (confirmed statistically)
- Most vulnerable ZIPs: 60608, 60609, 60629, 60644 (stat sig spikes, high Risk Scores)
- Mid-century homes (1940–70s) have highest lead levels
- HERZL Elementary flagged for extreme seasonal sensitivity
- Targeted Replacements: Focus on mid-century homes and high-Risk ZIPs
- Preemptive Summer Alerts: Launch campaigns in May
- Hotspot Monitoring: Prioritize statistically validated ZIPs (Risk Score > 1.5)
- Grey Zones: Expand sampling in ZIPs with insufficient test coverage
- Public Tooling: Use ZIP lookup for residents to check risk estimates
- Pranav Kuchibhotla: Temp-driven trend analysis, bootstrapping, animated maps, repo + visual design
- Obaid Bin-Mahfoudh: FOIA coordination, risk scoring model, street-level analysis, cost model
- Sebastian Buzenas: ArcGIS visualizations, hotspot analysis, census data integration
- Gavin Coffer: Website (v0.dev), temperature testing, poster edits
- Thailer Simmons: SQL database setup, VM optimization, multithreaded geocoding
- Advisor: Prof. Robert Ellis, College of Computing, Illinois Institute of Technology
- Expert: Elin Betanzo, P.E., Safe Water Engineering
- Data Sources: FOIA Requests, US Census, ADI, City of Chicago, EPA, NOAA
MIT License — See LICENSE