GeoConvert: Creating that Spatial Relationship David Rawnsley Mimas, University of Manchester
Overview GeoConvert is an online geography matching and conversion tool for UK academics. GeoConvert allows registered users to obtain and manipulate complex geographical and postcode data in a straightforward way. A solution to a specific UK problem but it can be applied to any country.
The Problem Many types of geography in the UK Administrative Electoral Postal Health European Census Statistical
A Little UK Geography Counties, Districts, Unitary Authorities Strategic Health Authorities Parliamentary Constituencies 2005
Further Complication WalesScotland England UA ST Ward CAS Ward Ward Unitary Authority ST Electoral Division CAS Postcode Sector MSOA LSOA LLSOA Intermediate Zones English and Welsh Output Areas Scottish Output Areas Northern Irish Output Areas GOR District County Council Area Data Zones District Councils CAS Electoral Division Northern Ireland UK ST Postcode Sector
A Partial Solution The NSPD Produced by the Office for National Statistics A long list of UK Postcodes Also has other geographical zones Postcode metadata New version every quarter
Problems With The NSPD It’s huge – 1.37 Gb at last count It’s cumbersome – flat file with many rows or basic Access databaseflat file with many rowsbasic Access database Horrible variable names – ‘OSHLTHAU’ It keeps changing Very little support from ONS Why can’t someone tame it?
Our Hero At Last What is it? Online web based access to NSPD Range of functions Provides a framework for integrating data Regularly updated Help and support provided FAQs Original documentation Glossary User Guides
GeoConvert
What Does It Actually Do? 3 functions Postcode related information Geography matching Geographical data conversion
Why Should I Use It? Prizes! It enables datasets to be linked much more easily It enables researchers to link their own data to many more datasets It enables data to be linked historically It allows users to quickly and easily get access to widely used deprivation and classification data
How Does It Do It? Postcode information direct from NSPD Use postcodes as a population proxy Dissection and re-aggregation of filtered postcodes between zones to achieve measure of proportional distribution of population Lots of complicated database stuff Lots of complicated interface stuff UK specific but the solution can be applied to any country
Isn’t Some Of That A Bit Dodgy? Dependent on accuracy of ‘how much?’ and ‘where?’ Boundary fuzziness due to unique allocation of postcodes Ideal data is population related with homogenous distribution male, female = good clusters of cancer patients = bad Based on current population current to historical works best Partial input = partial output
What Might It Do Next? A better population proxy? Different population proxies More metadata Confidence measures Standards based XML output Integration with external data sources Machine to machine Pipetastic