Dear colleagues,
Many helpful replies, received with gratitude! Stas
Kolenikov gave a really nice summary of the process. Two
other places I found useful ideas were Simon Ruggles
"Confessions of a micro-simulator" at
http://www.hist.umn.edu/~ruggles/Articles/confesssions of a
microsimulator.pdf and great technical report by Martin
Spielauer at
http://www.iiasa.ac.at/Admin/PUB/Documents/IR-02-032.pdf
Best wishes,
Anthony Staines
--
John `Mac' McDonald <[log in to unmask]>
Obesity e-Lab research project which started in September
2008 https://www.nibhi.org.uk/obesityelab/default.aspx
SIMIAN (Simulation Innovation: a Node) project a node of the
National Research Methods Centre <http://www.ncrm.ac.uk/>.
--
Simon Moore <[log in to unmask]>
You might find agent-based models interesting. There is also
a book called Socionics by Fischer et al that might be
relevant to your interests. Other authors to look out for
include Axelrod and Agar.
--
Ted Harding <[log in to unmask]> and Martin
Serpell <[log in to unmask]> both suggested
"agent-based modelling" as a very general computational
framework into which your application (as descibed) should fit.
--
Stas Kilenikov <[log in to unmask]>
You need to have a behavioral model that explains, in your
case, why and how people gain weight. It would probably
include expenses on healthy and not-so-healthy food,
exercise, commuting patterns and daily activities. In the
end, you will have something like regressions for daily
caloric intake and amount of calories actually spent as
dependent variables, and [fill the blanks] as explanatory
variables.
Then by varying your policy parameters, you will change
either the coefficients in the regression, or prices in the
demand for food, or incentives for exercise, or incentives
to bike to work rather than spend an hour in the car, one way.
I have worked on microsimulations for a tax model and for a
labor allocation model; there, the behavioral models were
quite simple, if not naive. How one can set up a model for
exercise and such, I have no clue.
--
On 07/10/10 16:34, Anthony Staines wrote:
> Dear colleagues,
> I am thinking about a piece of work, looking at the costs of
> different policies for managing obesity, for which
> microsimulation might be a useful option.
> A good snappy definition of microsimulation is
> "Microsimulation models are computer models that operate at
> the level of the individual behavioural entity, such as a
> person, family, or firm. Such models simulate large
> representative populations of these low-level entities in
> order to draw conclusions that apply to higher levels of
> aggregation such as an entire country. This type of model is
> distinct from aggregate models whose explanatory variables
> already represent collective properties." Statistics Canada
> http://www.statcan.gc.ca/microsimulation/index-eng.htm.
> In my case the "entities simulated" will be people. I am
> finding it hard to locate material on the nuts and bolts of
> actually doing microsimulation. There are lots of existing
> models for simulating specific stuff in specific countries,
> and there are lots of papers reporting the results of such
> analyses, but I have found neither :-
> 1) a flexible general purpose model framework which I can
> recycle,
> nor
> 2) a detailed exposition of how to set up such models from
> scratch and what can go wrong.
> Doubtless this reflects my ignorance. However, can anyone
> give me some pointers to either of these? I will recycle
> your summarized answers to the list.
> Best wishes,
> Anthony
--
Anthony Staines, Professor of Health Systems Research,
School of Nursing, Dublin City University, Dublin 9,Ireland.
Tel:- +353 1 700 7807. Mobile:- +353 86 606 9713
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|