SERVING THE QUANTITATIVE FINANCE COMMUNITY

 
User avatar
Cuchulainn
Posts: 59684
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Peak Life passed

April 3rd, 2019, 2:01 pm

 
User avatar
tw
Posts: 1054
Joined: May 10th, 2002, 3:30 pm

Re: Peak Life passed

April 4th, 2019, 8:30 am

Thank you so much for your long reply. 
You are very generous with your expertise and time, writing such a lengthy account with such a full reading list.

I scanned through some of the reviews, is it essentially Monte Carlo for social/local units with ad hoc/exogenous transition probabilities?
Is there any tie-in with ODE style approaches to population dynamics  birth-death process,predator prey or similar?


Fascinating stuff! Is there a good introduction you could recommend for how these microsimulations work?
Microsimulations are a broad set of methods and techniques for modelling micro- and large-scale dynamics of systems with a large number of units, which have different individual static and dynamic properties or interact with each other in non-trivial ways. They can be basic flow/traffic models or (not completely basic) epidemic simulations or vewy big models aggregating all data and facts about a studied system to be used for simulations and forecasts. You can use them to e.g.
- difficult: design an efficient country-size infrastructure for electric cars including all sorts of demographic and socioeconomic trends, measured traffic intensity at different times, etc. to optimally (according to a similarly complex utility function) allocate charging stations or funds
- simple: plan a health service's spending on treating diabetes based on the observed trends (population ageing, growing obesity in some groups vs. healthy lifestyle in others, ...)

You typically need such modelling most when you have enough information to model the system at the level of individual units or when your units have non-trivial "life histories" (e.g. a woman generally can't give birth twice in 9 months) or interact with each other (e.g. a child migrates with its mother; family members are likely to share similar diet or party inherit tendencies to obesity). One usually wants to look in such details in planning and management: 1-2% difference in can produce some critical effect or make a real difference if it accounts for several thousand people on the government healthcare bill - microsimulations can provide such precise predictions.

The modelling philosophy is use what you know and can to achieve the goal (I even went so far as to use deep learning, as you saw!). I've also never made a model which wouldn't include some level of agent-based modelling, like in this project (which some may say is more agent-based than microsimulation). : microsimulation of a botnet. And of course since it's a craft, not an art, it needs to serve some useful purpose: e.g. in the botnet project we use it to recover the actual structure of an operating botnet based on an observed stream of data the botnet sends to one of its computers (which was taken over by cybersecurity teams). In this way we know what the size and structure of the botnet is, and hence what threat it poses.

That's the gist of what's currently going on in this area (from my perspective).

I almost forgot about the references you've asked about - probably hoping that I wouldn't start to ramble. Sorry! :-) History, examples of models and applications:
G. Orcutt. A new type of socio-economic system. Review of Economics and Statistics, 58:773-797, 1957.
E. Morand, L. Toulemon, S. Pennec, R. Baggio, and F. Billari. Demographic modelling: the state of the art. SustainCity Working Paper, 2.1a, Ined, Paris., 2010.
E. Zagheni. Microsimulation in Demographic Research, volume 15 of International Encyclopedia of the Social and Behavioral Sciences, chapter B138, pages 343-346. Elsevier, 2nd edition, 2015.
L. Andreassen. Demographic forecasting with a dynamic stochastic microsimulation model.Technical Report 85, Central Bureau of Statistics, 1993.
G. Orcutt, J. Merz, and H. Quinke, editors, Microanalytic simulation models to support social and financial policy, North-Holland,Amsterdam, 1986.
E. A. Hammel, D. Hutchinson, K. Wachter, R. Lundy, and R. Deuel. The SOCSIM demographic-sociological microsimulation program operating manual. Technical Report 27, University of California, 1976.
M. C. Wolfson. Homemaker pensions and lifetime redistribution. Review of Income and Wealth, 34:221-250, 1988.
J. H. M. Nelissen and A. P. Vossen. Projecting household dynamics - a scenario-based approach. European Journal of Population, 5:253-279, 1989.
J. H. M. Nelissen. Household and education projections by means of a microsimulation model.Economic Modelling, 9:480-510, 1991

If you want to use microsimulations, I'd advise writing or buying a well-documented quality code. I wouldn't trust results of proprietary models, unclear methods or messy coding. It's generally not easy to implement a sensible microsimulation, but very easy to make mistakes. I've seen scary things in the market and some of them sadly went through peer-review in serious journals or serious people used them often paying inflated prices :-(
 
User avatar
katastrofa
Posts: 8038
Joined: August 16th, 2007, 5:36 am
Location: Alpha Centauri

Re: Peak Life passed

April 4th, 2019, 7:35 pm

My pleasure, tw. I got to the point where I produce such ramblings in minutes - not that I wouldn't happily use more time to answer your questions :-)

As you wrote, microsimulation is based on a stochastic simulation and the transition probabilities are calibrated to data and facts.

A simple (Markovian) ODE can't describe history-dependant dynamics and becomes inefficient for large systems - you could obviously use differential-integral equations and SDE to overcome these problems, respectively. However, it would be hard to model a system of a large number of units, each of which is different - which is basically what microsimulations do.
I would say a philosophical difference between the ODE approach and microsimulation is the source of this stochasticity: in the ODE approach it's a stochastic noise (a very large number of degrees of freedom we would never be able to model individually), while in microsimulations it's due to limited information about the system. That's why microsimulation has a status of a tool for statistical inference and is supposed to yield information about the uncertainty of results. I've seen nice ODE models of heart dynamics, as a comparison.
Another practical difference which comes to my mind is that in microsimulatons it's easier to modify the system arbitrarily (e.g. interventions), while an ODE can be less flexible.

Thank you so much for your long reply. 
You are very generous with your expertise and time, writing such a lengthy account with such a full reading list.

I scanned through some of the reviews, is it essentially Monte Carlo for social/local units with ad hoc/exogenous transition probabilities?
Is there any tie-in with ODE style approaches to population dynamics  birth-death process,predator prey or similar?


Fascinating stuff! Is there a good introduction you could recommend for how these microsimulations work?
Microsimulations are a broad set of methods and techniques for modelling micro- and large-scale dynamics of systems with a large number of units, which have different individual static and dynamic properties or interact with each other in non-trivial ways. They can be basic flow/traffic models or (not completely basic) epidemic simulations or vewy big models aggregating all data and facts about a studied system to be used for simulations and forecasts. You can use them to e.g.
- difficult: design an efficient country-size infrastructure for electric cars including all sorts of demographic and socioeconomic trends, measured traffic intensity at different times, etc. to optimally (according to a similarly complex utility function) allocate charging stations or funds
- simple: plan a health service's spending on treating diabetes based on the observed trends (population ageing, growing obesity in some groups vs. healthy lifestyle in others, ...)

You typically need such modelling most when you have enough information to model the system at the level of individual units or when your units have non-trivial "life histories" (e.g. a woman generally can't give birth twice in 9 months) or interact with each other (e.g. a child migrates with its mother; family members are likely to share similar diet or party inherit tendencies to obesity). One usually wants to look in such details in planning and management: 1-2% difference in can produce some critical effect or make a real difference if it accounts for several thousand people on the government healthcare bill - microsimulations can provide such precise predictions.

The modelling philosophy is use what you know and can to achieve the goal (I even went so far as to use deep learning, as you saw!). I've also never made a model which wouldn't include some level of agent-based modelling, like in this project (which some may say is more agent-based than microsimulation). : microsimulation of a botnet. And of course since it's a craft, not an art, it needs to serve some useful purpose: e.g. in the botnet project we use it to recover the actual structure of an operating botnet based on an observed stream of data the botnet sends to one of its computers (which was taken over by cybersecurity teams). In this way we know what the size and structure of the botnet is, and hence what threat it poses.

That's the gist of what's currently going on in this area (from my perspective).

I almost forgot about the references you've asked about - probably hoping that I wouldn't start to ramble. Sorry! :-) History, examples of models and applications:
G. Orcutt. A new type of socio-economic system. Review of Economics and Statistics, 58:773-797, 1957.
E. Morand, L. Toulemon, S. Pennec, R. Baggio, and F. Billari. Demographic modelling: the state of the art. SustainCity Working Paper, 2.1a, Ined, Paris., 2010.
E. Zagheni. Microsimulation in Demographic Research, volume 15 of International Encyclopedia of the Social and Behavioral Sciences, chapter B138, pages 343-346. Elsevier, 2nd edition, 2015.
L. Andreassen. Demographic forecasting with a dynamic stochastic microsimulation model.Technical Report 85, Central Bureau of Statistics, 1993.
G. Orcutt, J. Merz, and H. Quinke, editors, Microanalytic simulation models to support social and financial policy, North-Holland,Amsterdam, 1986.
E. A. Hammel, D. Hutchinson, K. Wachter, R. Lundy, and R. Deuel. The SOCSIM demographic-sociological microsimulation program operating manual. Technical Report 27, University of California, 1976.
M. C. Wolfson. Homemaker pensions and lifetime redistribution. Review of Income and Wealth, 34:221-250, 1988.
J. H. M. Nelissen and A. P. Vossen. Projecting household dynamics - a scenario-based approach. European Journal of Population, 5:253-279, 1989.
J. H. M. Nelissen. Household and education projections by means of a microsimulation model.Economic Modelling, 9:480-510, 1991

If you want to use microsimulations, I'd advise writing or buying a well-documented quality code. I wouldn't trust results of proprietary models, unclear methods or messy coding. It's generally not easy to implement a sensible microsimulation, but very easy to make mistakes. I've seen scary things in the market and some of them sadly went through peer-review in serious journals or serious people used them often paying inflated prices :-(
 
User avatar
Collector
Topic Author
Posts: 4217
Joined: August 21st, 2001, 12:37 pm

Re: Peak Life passed

August 2nd, 2019, 11:07 am

UK: Why is life expectancy faltering?

"For the first time in 100 years, Britons are dying earlier. The UK now has the worst health trends in western Europe – and doctors and experts believe that the impact of austerity is a major factor"....

"Our faltering life expectancy rates show we have now got the worst trend in health anywhere in western Europe since the second world war. To achieve that, we must have made a lot of bad decisions,” he said.

US life expectancy has been declining. Here’s why

phhuu, they need to think deeper than this...surface "science"
 
User avatar
Cuchulainn
Posts: 59684
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Peak Life passed

August 2nd, 2019, 1:47 pm

Junk food (e.g. couch potato), too much Mac, coffee, indolence, traffic jams,  opiates, forums, no social interactions, alcohol, cocaine, sugar, salt, Peroni, bitterballen, chicory, preservatives, whale meat, dull movies, The Smaiths, U2.

You would think that life expectancy would be longer.
 
User avatar
Cuchulainn
Posts: 59684
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Peak Life passed

August 2nd, 2019, 2:23 pm

“Nurse: “Doctor, the man you just gave a clean bill of health to dropped dead right as he was leaving the office”. Doctor: “Turn him around, make it look like he was walking in.””

— Henny Youngman

Image
 
User avatar
katastrofa
Posts: 8038
Joined: August 16th, 2007, 5:36 am
Location: Alpha Centauri

Re: Peak Life passed

August 2nd, 2019, 2:35 pm

If you eat a healthy plant-based diet, coffee works like the elixir of youth, but if you eat whales - not so much. Karma.
It's all individual: I always do what makes me feel good and don't care what others think about it.

The "Smaiths"? Please. You listen to The Corrs...

Re life expectancy in the UK - it's naive to analyse it without taking into account the migration patters. Besides, the old results are based on very small samples (large error), and likely overestimate the numbers owing to the survivor bias.
 
User avatar
Cuchulainn
Posts: 59684
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Peak Life passed

August 2nd, 2019, 3:14 pm

The "Smaiths"? Please. You listen to The Corrs...
Andrea is lovely and one of my sister's best friends, and Sharon took music lessons from my mum years ago.

www.youtube.com/watch?v=GJUVxMt1_W8
 
User avatar
katastrofa
Posts: 8038
Joined: August 16th, 2007, 5:36 am
Location: Alpha Centauri

Re: Peak Life passed

August 2nd, 2019, 5:45 pm

I'm sure Morrissey has some friends too...
 
User avatar
Cuchulainn
Posts: 59684
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Peak Life passed

August 3rd, 2019, 12:15 pm

I'm sure Morrissey has some friends too...
BNP?
https://en.wikipedia.org/wiki/For_Britain_Movement

and the Irish diaspora
https://www.thejournal.ie/edl-leader-ir ... 2-Jun2013/
 
User avatar
katastrofa
Posts: 8038
Joined: August 16th, 2007, 5:36 am
Location: Alpha Centauri

Re: Peak Life passed

August 3rd, 2019, 1:11 pm

Yeah, but you know, Meat is (a bigger) Murder.
 
JamesEnema
Posts: 2
Joined: August 4th, 2019, 12:40 am

Gape at is looking in the interest of a transgender actress also in behalf of the situation of a superhero

August 4th, 2019, 9:28 pm

https://usa.lgbt/study-lgb-teens-avoid-team-sports/
https://usa.lgbt/michael-jackson-the-last-trick/
https://usa.lgbt/spdqueer-boss-no-coali ... e-for-all/
https://usa.lgbt/award-for-queer-exchange/
https://usa.lgbt/when-the-big-love-retu ... -15-years/

Be awed Studios began the search as a replacement for a transgender actress who would reify the inception transgender seal in the old hat of the universe. The standard-bearer in the early casting inject was given the incognito “Jessica” and is to arrive in the fog to which the works choice start in 2020.

It seems that Miracle has conclusively begun to fulfill the promises made to the fans to spread the number of non-heteronormative heroes. The president of the studio, Kavin Faigh, said in June this year: “We are not hiding that this resolution materialize and that many more significant heroes and heroes of LGBT will become visible in the future.”

This all at once, the search in the interest of a transgender actress began with the original production. The happy of the proclamation reads: “A transgender actress age-old 20 to 30 with any ethnic unseen is sought as a replacement for the role.” Fans and fans speculate that she could be embodied in the forge of Sera – a lady associated with the the public of Thor and the most standard representative of trans people in the universe.
 
User avatar
Cuchulainn
Posts: 59684
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Peak Life passed

August 5th, 2019, 5:46 pm

Translation "I told you that I was sick".

Image
ABOUT WILMOTT

PW by JB

Wilmott.com has been "Serving the Quantitative Finance Community" since 2001. Continued...


Twitter LinkedIn Instagram

JOBS BOARD

JOBS BOARD

Looking for a quant job, risk, algo trading,...? Browse jobs here...


GZIP: On