search

divilian / specstar

Combines SPECscape and SPECnet into one project
1 stars 1 forks source link

Thinking about plots for the manuscript #8

Closed venkatachalapathy closed 5 years ago

venkatachalapathy commented 5 years ago

So, what are the plots that we might end of including in the manuscript? It depends on the scientific questions we are going to ask but thinking in terms of plots might help us ask more focused questions.

One 3D plot I had in mind was histogram of size distribution of proto-institutions versus network characteristics

Another one is the wealth distribution at the end point of our simulation versus the network characteristics

Anything else?

divilian commented 5 years ago

I'm a big fan of "Gini coefficient over time" for a single run of the simulation, with many overlapping line plots on different runs with the same parameters to show general trends.

When you say "size distribution of proto-institutions," do you mean a histogram that shows how many protos have exactly 2 members, how many have 3, how many have 4, etc.? Yes, that could be cool.

In general, I shy away from 3d plots because I feel that most humans (including myself) have difficulty meaningfully interpreting them. I feel their main value-add is to look cool.

For "wealth distribution at the end point of our sim" do you mean the entire Lorentz curve? Or just a Gini?

Pinging @will-nehrboss on this convo as well since we'll both probably be coding it.

wkneh commented 5 years ago

Once you nail down what this will be, send me an email to let me know what aspects I can build/code. Look forward to it.

Will

On Mon, Apr 15, 2019 at 12:16 PM Stephen Davies notifications@github.com wrote:

I'm a big fan of "Gini coefficient over time" for a single run of the simulation, with many overlapping line plots on different runs with the same parameters to show general trends.

When you say "size distribution of proto-institutions," do you mean a histogram that shows how many protos have exactly 2 members, how many have 3, how many have 4, etc.? Yes, that could be cool.

In general, I shy away from 3d plots because I feel that most humans (including myself) have difficulty meaningfully interpreting them. I feel their main value-add is to look cool.

For "wealth distribution at the end point of our sim" do you mean the entire Lorentz curve? Or just a Gini?

Pinging @will-nehrboss https://github.com/will-nehrboss on this convo as well since we'll both probably be coding it.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/WheezePuppet/specnet/issues/27#issuecomment-483319650, or mute the thread https://github.com/notifications/unsubscribe-auth/AvHrhChDoXTy9jDB8zQL6byR47FmtgH1ks5vhKXegaJpZM4cuYYL .

venkatachalapathy commented 5 years ago

Yes, I was thinking more in terms of a heat map where the z-axes is dependent on (x,y) and colored. And also some kind of histogram comparision test to find out if the any two interesting histograms are statistically different from one another.

divilian commented 5 years ago

So now that we have an animated wealth-distribution-over-time histogram, I'm realizing that doesn't exactly work for the manuscript. :) So I guess I should build a 2d one after all. Create new issue #28 for this.

venkatachalapathy commented 5 years ago

Since we have several issues open on this matter, this one feels like a redundant issue at this point.