Offender Modeling

This is an idea to create an agent based model for offending rates

some of the ideas are that .25% of all prisoners have been in care at some time compared to x% of the population

offenders profiles for economics offender profiles for education offender profile for health

health outcomes for offenders

Eclipse Agent-Based Modeling platform
AMP

I looked at some agent based models while studying for M366 Artificial intelligence course with the OU.

After listening to the "Bishop and the prisoner" radio 4 series (still not listened to the last one) I thought I would have a crack at agent modeling offending behaviour

I had previously noticed that eclipse have an Agent Based Modeling tool called AMP. Hence thought I would see if I could come up with a model to capture some of the ideas mentioned in the radio documentary, and that I have heard expressed

Ideas

1) criminals grow-out of offending behaviour, (rather than are rehabilitated)

2) imprisonment is cheaper than the criminal behaviour(civitas)

3) imprisonment is not cheaper than community solutions( ken Clarke)

4) UK life sentences seem to be 10-15 years, in the US, life means life.

5) what types of outcome effects could be produced by various parameter changes

Some criminal characteristics are time-based, ie type of offenses change over time, severity of offences

some agent variables
Education

Poverty

Health

some systems variables
Incarceration length

police officer numbers

probation officer counts

some interactions/dynamics to be modelled
probation?

parenting?

also, what about setting some outcomes, and monti-carloeing for plausible population variable values

if murder is say 0.01%, what does that imply about various social parameters

$$ \sum f(x) = F(x) + g(x) $$