Machine Learning in Public Infrastructure

Public infrastructure is a demanding setting for machine learning because predictions often mediate access to scarce inspection, repair, or response capacity.

Examples from the unpublished blog drafts:

• finding lead service lines in Flint
• prioritizing review of sewer pipe inspection video
• forecasting gas odor calls and response-center staffing
• using excavation permits to route watch-and-protect teams

These applications share three constraints:

1. They are resource allocation problems.
2. They depend on domain knowledge.
3. They produce consequential predictions.

This cluster is the working substrate for What Utilities Taught Me About Machine Learning.