Andy Turner's MoSeS Meeting 2006-05-09 Page

Andy Turner's MoSeS Home Page

People

• Mark
• Phil
• Martin
• Andy

Ice Breaker

• Mark returned from San Diego and told us of a few folk he met including the EPISIMS people...

Documentation

• Mark brought a draft of the model specification he is working on, but this was not distributed.
• Phil circulated print out of the various comments on notes from MoSeS meeting 2006-04-12.
• Both Andy and Phil made notes during the meeting.

Upcoming meetings

• e-Social Science Conference - last week in June
• Information Centre
  • Denise - data archive, Martin, Mark
  • Health and social services
  • Seminar 9th June
  • Information Centre keen to set up Masters projects - work placements
  • Myles Gould seconded Information Centre 3 days per week
• Research Methods Fest - July 18-21st Oxford
• Supercomputing conference in Florida (November)
  • EPSRC Proposal needs developing by the end of May
    • Rob Proctor is keen for MoSeS to organise something on behalf of NCeSS, so this is somewhat of a must...
    • 3 day demonstration sounds like hard work...
  • MoSeS Meeting Leeds Primary Care Trust second meeting 2006-06-06

Notes

• How many Communal Establishments are there in Leeds?
• Semantic confusion
  • What Andy calls "constrain variables" Phil has termed "control variables"
  • What Andy calls "optimisation variables" Phil has termed "constraint variables"
• Jin has completed his IPF code. This is being written up as a paper...
• Currently the MoSeS demographic model is developing to split Communal Establishment (CE) from Household populations and using the ISAR
  • Originally CE populations were not dealt with separately which caused problems where there was a mix of CE and Household populations in an Output Area
  • In the next iteration the MoSeS demographic model will try using HHSAR instead of ISAR for households
  • The MoSeS demographic model may in the end use both ISAR and HHSAR for constructing households
• Martin's way of doing population reconstruction would be to start with a household and then populate it with individuals. This is a good way and another way of doing it that may be complimentary.
• Jin and Phil have generated IPF probability matrix to create individual ages from aggregate ages
  • Potentially very useful if HHSAR is to be used (due to its poor Age resolution)...
• What is meant be a family in the census? What are the definitions of multi-family households?
• Discussed the difficulties in constraining by age using HHSAR data...
• Phil's population reconstruction method
  • Select HRPs base number of households and HRP age on CAS003 table
  • Optimise to get right number of people and the right age breakdown
  • This is yet another way of doing it that should be done and results and method compared with other ways...
• The major problem with using the HHSAR is that it is only England and Wales.

• No special license required
• Spatial resoultion of LAD?
• Jin is looking at the data in detail
• Andy had a quick look at the data and it seems many of the important variables are there...