Skip to content
chris48s edited this page Dec 18, 2017 · 3 revisions

UPRNs as a first-class Citizen in WhereDIV: 1 - Motivations

Why did we want to do this, what problems did we hope to solve?

  • Remove 'blacklist' + provide results on the edges of local auths
  • Detect and solve certain classes of errors in council data
  • Simplify edge case handling
  • Improve performance: everything is a PK lookup
  • Reason about performance consistently: performance will be roughly the same with 1% coverage and 100% coverage (using denormalised model)
  • Know that we are not making incorrect assumptions about postcodes based on incomplete data
  • Certain classes of error become much easier to detect (e.g: polling district doesn't quite clip to local auth boundary)
  • Gets you towards a case where the user journey can be "start typing your address..."
  • If the council supplies an ambiguous address but it has a UPRN attached, we can discard less ambiguous data and achieve better coverage rates
  • We could query EE by point to avoid 'fuzzy edges'