EDA of economies in colonized systems

[CMDR kekosummer]

Was interested in overall state of economies that turned out.
Filtered only stations that were created by players (id begins with 42...) , excluding construction sites, updated after TB3 rolled out.
Data source: Spansh stations dump

Notebook published on Kaggle: https://www.kaggle.com/code/iananic...ed-systems/notebook?scriptVersionId=241413622
It has more plots then shown here. Also, all plots are interactive, and can be downloaded in a single zip to explore further.

Value of 1 is equivalent to 100%. So for example, Coriolis in orbit of HMC with pristine resources will get 140% == 1.4 extraction economy.

Global sum of economy proportions

Split by station type

Box plot of distribution proportions

Split by station type

Distribution of stations per economy proportion
(here I only show 5 economies - more in the notebook; log10 scale on Y axis)

 

[CMDR kekosummer]

Updated: https://www.kaggle.com/code/iananic...f-colonised-systems?scriptVersionId=242499469

Added summary of station type counts:

(yeah, I am far from understanding why so few planetary ports displayed here)

And similarly to the grouped histogram with eceonomy first, station type second, here's reversed grouping:

Added top 20 station rankings per economy type.
Example below:

 

[Ned Flandalorian]

This would only be based on third-party data collection (e.g. EDDN) right? I'd say it's missing the majority of colonized systems.

 

[CMDR kekosummer]

This would only be based on third-party data collection (e.g. EDDN) right? I'd say it's missing the majority of colonized systems.

Is it missing many? There's more then 20 000 systems here, and about 25 000 stations. Not for from actual numbers in my understanding.
It's enough for anyone who sends data to EDDN to simply visit system and stations' economy data would be recorded.

Anyway, I believe it will represent the active core of colony stations.

 

[Ian Doncaster]

This would only be based on third-party data collection (e.g. EDDN) right? I'd say it's missing the majority of colonized systems.

From Frontier's earlier official data releases, about 2/3 of colonised systems had got an EDDN ping reflecting that status (and of those systems, most of them have data that's only 1-2 weeks old). That ratio has probably improved in the months since as the pace of new system addition slows down.

So I'd expect the majority of systems to be in it - especially now that the rate of colonisation has slowed down significantly - but it's very likely that some of the station types (settlements, most obviously) are under-represented because a lot of them will have been built for their system properties, rather than to dock at.

 

[Ned Flandalorian]

I mean I can pretty much pick any colonized system around me, look it up on Inara and the details are wrong; missing stations, missing facilities, out-of-date economy info due to the latter...

 

[Ian Doncaster]

I mean I can pretty much pick any colonized system around me, look it up on Inara and the details are wrong; missing stations, missing facilities, out-of-date economy info due to the latter...

Certainly. But for questions like what's the proportion of assets of a particular type / economy / etc. it might still be about right considering all known systems.

(Except for settlements, where it'll be massively underestimating their numbers, though probably not too bad on type)

 

[Astrofizzix]

I mean I can pretty much pick any colonized system around me, look it up on Inara and the details are wrong; missing stations, missing facilities, out-of-date economy info due to the latter...

I’m inclined to agree. I’d take anything from Inara with a big grain of salt.

I probably screwed up the search, but I couldn’t find the system I’ve been working on the past six weeks in that database. And even Inara has something at least for it. Albeit with several of the locations still showing as construction sites in spite of being completed. Maybe I’ll try again when I’m at my computer to see if I missed something on my phone.

Col 285 Sector KI-A C14-8

 

[CMDR kekosummer]

I probably screwed up the search

Yeah that's definitely search from Kaggle.com, not Inara.cz, which ofcourse won't have that system

 

[Astrofizzix]

Yeah that's definitely search from Kaggle.com, not Inara.cz, which ofcourse won't have that system

lol. I’ll try again when I have a bigger screen. Thanks for confirming.

 

[CMDR kekosummer]

Another update:

Elite Dangerous: economy of colonised systems

Explore and run machine learning code with Kaggle Notebooks | Using data from No attached data sources

www.kaggle.com

 

[CMDR kekosummer]

New version:

Elite Dangerous: economy of colonised systems

Explore and run machine learning code with Kaggle Notebooks | Using data from No attached data sources

www.kaggle.com

(it just began download of fresh data so might take a while to complete)

What's new:

• extracting and saving as CSV market data (supply/demand of commodities)
• quick analysis of a limited selection of commodities' supply/demand relative to economy it's produced in

Examples:

It has 3 subplots:

• most left is a box plot that is good for distributions (except here in most cases it quickly collapses because most points are close to 0)
• next goes a simple histogram - number of stations per supply/demand bin
• and last is a density heatmap, with the distribution of supply/demand on a station vs economy proportion of that same station, split by tiles, each colored according to a number of stations that fall into each tile.

Separately built plots for all stations and only including stations with said economy strength greater or equal 1.

Drawing conclusions, it can be said, for example, that supply/demand of Insulating membranes is greatly unbalanced, without ever crossing 50kT supply in any of the colonised systems we have data on. Compared to much more balanced Tritium.

 

[Ian Doncaster]

Insulating membrane seems to be a fixed 0.3 tonnes per economic unit since Frontier boosted its supply, which means to get above 50kT on a single station you'd need a population-equivalent of 27 billion.

(Obviously no-one is even approaching 50kT that way - being able to put more than 1.0 Refinery fraction on a station, or boost supply above the NPC equivalent because development level seems to work differently in colonised systems, means that it could be done with a smaller market than that!)

and last is a density heatmap, with the distribution of supply/demand on a station vs economy proportion of that same station, split by tiles, each colored according to a number of stations that fall into each tile.

It would be interesting to see this with a sum of the economy proportions importing and exporting - e.g. for Insulating Membrane, rather than plotting Refinery, plot (Refinery) - (Industrial + Military + High-Tech)

 

[PilotB]

Is that a typo in the tritium chart or have you actually plotted extraction vs trit supply? As far as I understand it tritium is a refinery item not extraction despite our way of getting it being mining it not refining it.

 

[CMDR kekosummer]

Is that a typo in the tritium chart or have you actually plotted extraction vs trit supply? As far as I understand it tritium is a refinery item not extraction despite our way of getting it being mining it not refining it.

Well, plot doesn't have a typo - it is Tritium supply/demand vs refinery economy strength.
Why? Because I thought it is that was, since my mainly refinery surface outpost didn't produce it, while Coriolis with 140% did, and I got that impression from discussions on that topic. Should have checked properly...

I must say that this selection of commodities wasn't well thought out and served more as "something to test with quickly" If you want you can call that an alpha-version of the list, though plots must be solid, even if connection was made wrong.
Tritium is the only one I took from my head, the rest I took from https://cmdr-nowski.github.io/scuffed/

 

[PilotB]

deleted

 

[CMDR kekosummer]

Insulating membrane seems to be a fixed 0.3 tonnes per economic unit since Frontier boosted its supply, which means to get above 50kT on a single station you'd need a population-equivalent of 27 billion.

(Obviously no-one is even approaching 50kT that way - being able to put more than 1.0 Refinery fraction on a station, or boost supply above the NPC equivalent because development level seems to work differently in colonised systems, means that it could be done with a smaller market than that!)

Well, maybe we colonies don't need even 10kT supply (to build new ones and engage in some light trade), but it still looks very unbalanced. If there were any penalties for not meeting demand, this would be a disaster. Demand is huge!

Though maybe not as terrible as Beer: someone must be really very extremely thirsty

It would be interesting to see this with a sum of the economy proportions importing and exporting - e.g. for Insulating Membrane, rather than plotting Refinery, plot (Refinery) - (Industrial + Military + High-Tech)

I thought of something similar.
I don't know where I could easily (without tedious manual work) grab table of commodity, produced_by, requested_by to make it simple...
So my idea was to just separately plot stations where plotted economy has more then 80% of total strength (aka not flat >=0.8 but >=0.8*sum(economies) ). And for now (that latest version linked above) just got separate plots for when strength exceeds certain limit.

 

[Ian Doncaster]

If there were any penalties for not meeting demand, this would be a disaster. Demand is huge!

Yes - it's not set up to be a balanced economy where player trading forms a measurable fraction of supply/demand regeneration rates.

Fortunately it doesn't need to be. The pre-colonisation bubble had 100 billion tonnes of goods on the market, regenerating at probably every 2-3 days or so on average, so requiring >10 million player-hours a day of maxed T-9/Cutter/Carrier hauling to exhaust it.
(Or in more comprehensible units, maybe 500-2000 times the hauling rate at a reasonably popular trade CG)

 

[CMDR kekosummer]

It would be interesting to see this with a sum of the economy proportions importing and exporting - e.g. for Insulating Membrane, rather than plotting Refinery, plot (Refinery) - (Industrial + Military + High-Tech)

Forgot to ask - is it fair to treat supply of economy A against demand of economy B as equal units?
I got impression that demand if just a single weak link can be enough to severely decrease supply of other (major) economies.

 

[Ian Doncaster]

Forgot to ask - is it fair to treat supply of economy A against demand of economy B as equal units?

That appears to be the case in general, but it would be worth more rigorous testing if you can easily pull the data together.

The way I'd construct the test is:
- pick a single commodity of interest
- firstly, find the single-type economies of types which either import or export it
- get the range of import/export levels, calibrating them by measuring not in tonnes but as a percentage of the Hydrogen Fuel export tonnage at the same station
- that should get two (separate) distributions of import and export strength relative to the HFuel values
- from which it should then be possible to do check if the supply/demand of mixed economies meets that hypothesis

I got impression that demand if just a single weak link can be enough to severely decrease supply of other (major) economies.

Yes. But equally a single weak link can do the same to demand, for commodities which have heavy supply-bias. Progenitor Cells - export HT, import everything else - might provide a good test case, as you can get the export fraction as a simple "HT/Total" calculation, then correlate that to the supply or demand.

(The effects on demand can be equally large, but players are far less concerned - for now, anyway! - about what colonisation stations buy than about what they sell, because they're not making the stations to sell to, and the commodities they buy there probably aren't going to be sold to a market either)