Finding Evidence in Water-Service Records
OCR helped find water-service records for checking possible lead-pipe-free locations.
Why it matters
Searching historical drawings can surface service-line evidence that is difficult to find manually. Potential matches give staff records to review; they do not confirm pipe material.
My contribution
I wrote the Python and R workflows that read text from historical drawings and flag possible lead-free locations for staff review. The location count, savings estimate, and award below belong to the wider team effort.
Result
The broader City effort identified 1,000+ potential lead-pipe-free locations, with $1M+ in estimated savings. The City team was a 2023 TCEQ Technical/Technology award finalist for Lead Services.
Tools
- Python
- R
- OCR
- Record analysis
Potential locations still need material confirmation, and the savings are estimates. The separate synthetic ML prototype does not show historical accuracy or realized savings.
Technical details
Professional · Geospatial Software Developer / Data Scientist
Useful service-line details were stored in historical drawings that were difficult to search.
- Read text from historical drawings with OCR.
- Search the extracted text and group related service records.
- Send possible matches for review against the service-line inventory.