ENH: Treatment of agent characteristic change in network submodel (CopanBehaveModel class))

[jdonges]

Agent characteristic change as described in the paper for network submodel is not implemented in CopanBehaveModel class. Quote from paper:

a \emph{network} model not considering local social influence, but inducing behavioural change only whenever the exogenously modified smoking disposition $\gamma_i(t)$ of an individual $i$ crosses a threshold of 0.5 (modifying step 3)

Where is this implemented? In some external run script?

I vote that for consistency this dynamics should be included in the model class instead (because all other submodels are clearly part of it as well now).

[carlschleussner]

@jdonges:
Sorry to be frank (don't take it wrongly), this is a kind of a comment that comes a bit late in the game.

We had some discussions previously that we would like to separate the actual behave model from the smoking implementation. We discussed this is some length and this is how it's done by and large now. So generic elements are in the CopanBehaveModel class, whereas additional and case specific implementations like the change in agent characteristics are described in bin/ens_smoking_transition.py
This is also explained in the readme.

Clearly, it's not that clear cut and even less so by all the changes we recently implemented to be consistent with the manuscript (e.g. adding specific agent characteristics and properties instead of a more general array-based approach). I think that's ok.

So this could be changed, obviously, but is a bit more complicated then just c&p. Given that we have a deadline in two days, that I spent several days on going over this a while ago (where there would have been a window to change this) and given that we may want to prioritize content work, I would vote against it.

Please check the bin/ens_smoking_transition.py. I hope you can agree to closing that issue.

[jdonges]

Problem is that I am still in the process of understanding the code holistically step by step.

I agree that for pragmatic reasons we should stop. OK. But I still think it is not the best division of code into class and run script then, but nevermind… It’s research code…

Am 28.06.2016 um 17:27 schrieb carlschleussner notifications@github.com:

@jdonges https://github.com/jdonges:
Sorry to be frank (don't take it wrongly), this is a kind of a comment that comes a bit late in the game.

We had some discussions previously that we would like to separate the actual behave model from the smoking implementation. We discussed this is some length and this is how it's done by and large now. So generic elements are in the CopanBehaveModel class, whereas additional and case specific implementations like the change in agent characteristics are described in bin/ens_smoking_transition.py
This is also explained in the readme.

Clearly, it's not that clear cut and even less so by all the changes we recently implemented to be consistent with the manuscript (e.g. adding specific agent characteristics and properties instead of a more general array-based approach). I think that's ok.

So this could be changed, obviously, but is a bit more complicated then just c&p. Given that we have a deadline in two days, that I spent several days on going over this a while ago (where there would have been a window to change this) and given that we may want to prioritize content work, I would vote against it.

Please check the bin/ens_smoking_transition.py. I hope you can agree to closing that issue.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub #17 (comment), or mute the thread https://github.com/notifications/unsubscribe/ADCbe3k-FOgYEIVTUkYOP0dgGTuvGAzaks5qQT1xgaJpZM4JANPv.

[carlschleussner]

Yes, I understand that and I may even agree. But the last time we discussed this very issue of code separation (which was about two weeks ago), we have arrived at the conclusion that we see merit in keeping it separated. So this is what I did and this is where we are now.

On 28 Jun 2016, at 17:42, jdonges notifications@github.com wrote:

Problem is that I am still in the process of understanding the code holistically step by step.

I agree that for pragmatic reasons we should stop. OK. But I still think it is not the best division of code into class and run script then, but nevermind… It’s research code…

Am 28.06.2016 um 17:27 schrieb carlschleussner notifications@github.com:

@jdonges https://github.com/jdonges:
Sorry to be frank (don't take it wrongly), this is a kind of a comment that comes a bit late in the game.

We had some discussions previously that we would like to separate the actual behave model from the smoking implementation. We discussed this is some length and this is how it's done by and large now. So generic elements are in the CopanBehaveModel class, whereas additional and case specific implementations like the change in agent characteristics are described in bin/ens_smoking_transition.py
This is also explained in the readme.

Clearly, it's not that clear cut and even less so by all the changes we recently implemented to be consistent with the manuscript (e.g. adding specific agent characteristics and properties instead of a more general array-based approach). I think that's ok.

So this could be changed, obviously, but is a bit more complicated then just c&p. Given that we have a deadline in two days, that I spent several days on going over this a while ago (where there would have been a window to change this) and given that we may want to prioritize content work, I would vote against it.

Please check the bin/ens_smoking_transition.py. I hope you can agree to closing that issue.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub #17 (comment), or mute the thread https://github.com/notifications/unsubscribe/ADCbe3k-FOgYEIVTUkYOP0dgGTuvGAzaks5qQT1xgaJpZM4JANPv.

—
You are receiving this because you were assigned.
Reply to this email directly, view it on GitHub #17 (comment), or mute the thread https://github.com/notifications/unsubscribe/ADCbhLnNHi6VLa541ZMkq6Hijxzxdyn5ks5qQUDagaJpZM4JANPv.

[dengemann]

Maybe it helps.

On Tue, 28 Jun 2016 at 17:48, carlschleussner notifications@github.com
wrote:

Yes, I understand that and I may even agree. But the last time we
discussed this very issue of code separation (which was about two weeks
ago), we have arrived at the conclusion that we see merit in keeping it
separated. So this is what I did and this is where we are now.

On 28 Jun 2016, at 17:42, jdonges notifications@github.com wrote:

Problem is that I am still in the process of understanding the code
holistically step by step.

I agree that for pragmatic reasons we should stop. OK. But I still think
it is not the best division of code into class and run script then, but
nevermind… It’s research code…

Am 28.06.2016 um 17:27 schrieb carlschleussner <
notifications@github.com>:

@jdonges https://github.com/jdonges:
Sorry to be frank (don't take it wrongly), this is a kind of a comment
that comes a bit late in the game.

We had some discussions previously that we would like to separate the
actual behave model from the smoking implementation. We discussed this is
some length and this is how it's done by and large now. So generic elements
are in the CopanBehaveModel class, whereas additional and case specific
implementations like the change in agent characteristics are described in
bin/ens_smoking_transition.py
This is also explained in the readme.

Clearly, it's not that clear cut and even less so by all the changes
we recently implemented to be consistent with the manuscript (e.g. adding
specific agent characteristics and properties instead of a more general
array-based approach). I think that's ok.

So this could be changed, obviously, but is a bit more complicated
then just c&p. Given that we have a deadline in two days, that I spent
several days on going over this a while ago (where there would have been a
window to change this) and given that we may want to prioritize content
work, I would vote against it.

Please check the bin/ens_smoking_transition.py. I hope you can agree
to closing that issue.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub <
https://github.com/pik-copan/pycopanbehave/issues/17#issuecomment-229085189>,
or mute the thread <
https://github.com/notifications/unsubscribe/ADCbe3k-FOgYEIVTUkYOP0dgGTuvGAzaks5qQT1xgaJpZM4JANPv
.

—
You are receiving this because you were assigned.
Reply to this email directly, view it on GitHub <
https://github.com/pik-copan/pycopanbehave/issues/17#issuecomment-229090027>,
or mute the thread <
https://github.com/notifications/unsubscribe/ADCbhLnNHi6VLa541ZMkq6Hijxzxdyn5ks5qQUDagaJpZM4JANPv
.

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#17 (comment),
or mute the thread
https://github.com/notifications/unsubscribe/AB0fitqEA5CtBQ6hHhcQaZ5Uhupvr4DSks5qQUJngaJpZM4JANPv
.

[jdonges]

Let's leave this as an optional improvement / refactoring for later.