Skip to main content

Recipes

Task-oriented walkthroughs for jobs people actually show up with: geocode a table, put the results on a map, keep a coordinate private. Each one statesregionThe first-level administrative subdivision of a country — a US state, a French region, a province. The component between country and locality. what you'll have by the end and links the reference or concept page it builds on.

  • Batch geocoding — turn a list of addresses into a list of coordinates with one bulk request, with per-row error isolation so one bad row never sinks the batchbatch sizeHow many examples the model processes before each parameter update. Larger batches give smoother gradients but cost more memory; gradient accumulation simulates a big batch on a small GPU..
  • Displaying results on a map — render resolved entities to a self-contained map.html you can open in a browser, from the same data you'd export as GeoJSON for QGIS.
  • The free first pass — run Mailwoman locally as a no-cost first geocodinggeocodingThe process of converting an address into geographic coordinates (latitude and longitude). Mailwoman geocodes in a multi-tier cascade: exact address-point match → street interpolation → locality centroid. Each tier is progressively coarser but more widely available. pass, and spend a paid API's budget only on the residual it couldn't pin well enough.
  • Ingesting giant CSVs — stream a multi-gigabyte CSV into normalized records on one threadthreadA parallel workstream within a release. Threads compose; they are not sequential milestones like phases., then threadthreadA parallel workstream within a release. Threads compose; they are not sequential milestones like phases. only the geocodinggeocodingThe process of converting an address into geographic coordinates (latitude and longitude). Mailwoman geocodes in a multi-tier cascade: exact address-point match → street interpolation → locality centroid. Each tier is progressively coarser but more widely available. stagestageOne of the dataflow stages in the runtime pipeline (normalize, locale gate, kind classify, phrase group, token classify, sequence correct, reconcile, resolve). Distinct from tier (model vocabulary) and phase (plan milestone). that's actually expensive.
  • Coarsening a coordinate for privacy — round or geohash a rooftoprooftopGeocoding precision at the building or parcel level — coordinates within a few metres — the highest tier of the geocode cascade. Sourced from address-point and situs data. coordinate down to neighbourhood-level precisionprecisionOf the spans the model labeled as a given tag, the fraction it got right. High precision means few false positives. Paired with recall to compute F1. before it lands in an analytics table or a public dashboard.
  • Looking up a timezone — resolve a coordinate to its IANA timezone over a self-built point-in-polygon database.
  • UN/LOCODE lookup — resolve a place nametoponymA proper name for a geographic place. or coordinate to its UN/LOCODE trade-and-transport code, and back.