Analysing the ELW data from EDDB/EDSM

[marx]

Posting here because I've started doing the final update to this, with today's data, and I don't want the thread auto-locked before I finish it. Should be ready by June 15, but best to make sure.

In the meantime, does anyone have some ideas on what would be good to include that wasn't before?

 

[marx]

Updated the ELW distribution a year later, as promised. See the updated first post for the link. There's more to come, but I still have some new stuff I want to check, and I can no longer add the per-ELW breakdown to Google Sheets, as it can't handle that much data. I'll probably paste the results there, and share the spreadsheet with the data separately.
As before, I recommend getting a local copy of the map to play around with (see the ratio cut-offs on the Data sheet), especially since there is more data now.
I'll make another sheet with the ringed ELWs separately displayed as well, as I did before; I just did it this way now because it loads Google Sheets down even more. (Plus I've misplaced my data, and it might take a few days until I can get it back.)

Note that I didn't include the Wregoe and Synuefe sums this time around, because I mistakenly counted some sectors twice there. The cause being that not all of the overrides fit neatly into just one sector. So, take the ratios near the bubble with a good spoonful of salt: a lot of ELWs are "lost" to the various spherical sectors there.


As for how things have progressed in the year, with twice as many ELWs found: to me, it looks more and more likely that ELWs might be slightly more prevalent inside the core. Even if that suspicion really is true, though, the difference in chances is still much less than it is if you filter by star classes.

Another interesting note is the rim. Even though the DECE completed since the last update, under the default cut-off values (minimum 10 ELWs in minimum 10000 systems) the rim doesn't even appear. If you look at the "plain" counts instead of the ratios though, it shows up nicely. Even under more generous limits though, there simply aren't enough systems in these sectors on the rim to conclude anything.
Not surprisingly though, this is only for the galactic plane: above and below layers 0 and -1 (the latter being courtesy of Colonia), ELW counts drop dramatically. Little wonder, of course.

 

[marx]

Never mind my earlier comment, I could complete and upload the ringed ELW version too. See the first post for the link.

 

[marx]

So, updated the first post with a link (here) for further analysis of ELWs. This time around, I couldn't upload all the data, just the results, the ringed ELW data and the coordinates: the reason being that Google Sheets couldn't handle that much data. If somebody would like to see the originals, pop me a PM and I'll send them over.

Thanks to Orvidius' EDAstro sheets on the EDSM data, this does include more things about systems and the galaxy than last year's analysis did. These would be:
1. ELW-bearing systems' main stars, solo stars, and ELWs' parent starts, compared to the total star count per type
Pretty much the same we knew before, but it's good to see them all here.

2. ELM parent body types
What kind of bodies Earth-like moons orbit. Not surprisingly, HMCPs make up 69% of this.
"MISSING" here means that the parent body wasn't scanned by the submitter, so it's unknown.

3. Exploration activity by month, via ELWs, systems and body totals added
Two curious things here. First, the release of Chapter Four with the easier and quicker scanning of ELWs didn't noticeably increase the ELW / Sys ratio, unlike the ELW / Bodies ratio. Second, the rate at which ELWs were discovered after DW2 began winding down dropped to less than half, despite roughly the same amount of systems discovered per month. I think the most likely scenario here is that people were hurrying back to the bubble, and probably using neutron stars en masse. Or just simply not even calling up the FSS, just jump-scoop(-honk-)jump.


Also, I didn't include anything on the planetary characteristics, for two reasons. One, I used Orvidius's compilation, and noticed only too late that it doesn't have the surface pressure listed there, and didn't want to compile my own by then. Two, I checked on all the other characteristics, and things were the same, nothing new emerged. I can still send the original sheet with the data over if anyone wants, minus the surface pressure of course.

 

[Orvidius]

Yeah, surface pressure was something that I overlooked when I first built the database. I have it all imported now, as of about a week ago. I can start adding it to the spreadsheets now.

EDIT: Assuming I didn't hose anything up, today's update should include columns for pressure.

 

[marx]

Sweet, thanks! I'll take a look at that after I get back from holiday, and post it here if there's anything new there.

Also, one minor thing I forgot. In the process of removing non-natural ELWs (those that were terraformed or are in non-proc. gen. systems), I accidentally removed those few as well which are in proc. gen. systems that were later renamed. (For example, Metztli.) This is a negligible percentage of the total though, but if someone's looking through the ringed data for a few of their favourite systems, that's why they aren't there.

 

[schlowi123]

I did the same plots as I did for the Ammonia worlds. It probably is the same as in spreadsheets but … øhm … I'm too stupid to work with spreadsheets. I prefer plain text files and run my algorithms over them directly.
Anyway, date of the source: 2019-06-18.
Also: JUST the data after the FSS was introduced was analyzdes

First the absolute numbers of ELW's found in systems with the given main star (omitting small quantities):

In comparison with the AW's the following is different:

• lower total numbers (of course)
• An ELW in a system with a M type main star is almost equally likely as in a system with a Neutron main star. For AW's the former was much higher.
• K (main) star systems have NOT any longer the largest count.
• The number of ELW's in F (main) star systems is just approx. twice as large as in A (main) star systems. This is important for the probabilities:

Second, the probability to find an ELW in a system with a given main star:

This is very different compared to the AW-results:

• The probabilities are of course smaller.
• It is more likely to find an ELW in an A (main) star system than in a F (main) star system (it was the other way around for AW's). However, the chance to find an AW or an ELW in an A (main) star system is approx. equal! That was a bit of a surprise
• It is more likely to find an ELW in a Neutron (main) star system than in both a K or M (main) star system. But again, the chance to find an ELW or an AW is approx.
• The largest surprise is that the chance to find an ELW in a M (main) star system is so small.

Edit: Error in the ordinate label of the second image. It is of course < Chance to find an ELW […] >.

 

[schlowi123]

As for AW's did I do the same for the (most common) masscodes / star type combinations.

Always writing "main" is tedious. So when I write "star" or such below, it is ALWAYS the main star of a system I refer to. As a reminder: this does NOT need to be the parent star the ELW is orbiting. But we can filter just for the former.

First the absolute numbers.

Compared to the AW's the following is different:

• The absolute numbers (of course).
• Not many ELW's in Neutron star systems.
• More ELW's in F-d systems than in K-c systems. For AW's it was the other way around. The difference is significant, where it was for AW's small.
• Double as many ELW's in A-d systems compared with G-c systems. For AW's the numbers where almost equal.

And now the probabilities:
(Edit) ATTENTION: Due to a copy and paste error are the probabilities for all M type stars wrong! At the very end of this post I present the actual probabilites but for the sake of the discussion I leave the old results here. This is valid JUST for the M type stars results. All other results are correct!

WHOA! We have a clear winner: If you want to find ELW's, go to systems with an M type main star (red dwarf that is) and masscode < e >. That came as a total surprise, especially after seeing the absolute numbers above. It may be an artifact though, since the numbers are 43 ELWs found in 430 M-e systems. on the other hand, the numbers are high enough that I would risk a bet. Then again … there seem to be not many M-e systems (red dwarf). So it will be a while before it can be decided if I win this bet or not … I would say I have a 35 % chance to win this.

So, let's not concentrate on this and look closer at the other combinations:

I don't trust the F-e, G-e, K-e results, since the number of ELW's found in these systems is 13 or smaller. The rest is OK
Compared with the AW's we have the following differences:

• The chances are in general smaller to find an ELW.
• M-b is not favourable for ELW's (while it has a number of AW's)
• In M-c and M-d systems it is more likely to find ELW's.
• M-d systems have a decent chance to have an ELW in them.
• It is more likely to find an ELW in an A-d systems than in an F-d systems (it was the other way around for AW's).
• It is just half as likely to find an ELW in an F-e systems than in an F-d systems (for AW's the difference was small)

So overall I would say that the chance to find an ELW is highest in A-d systems.
There aren't that many M-d systems ("just" a bit more than 45,000 in the data I used) to recommend looking for them. But if one find's such a system, the chances are almost as high as in A-d systems.
Second best "hunting ground" are clearly F-d systems. There are more than double as many of them as there are A-d systems so these are easier to find.

If looking for the masscode takes too much time: filter for A and F systems. If the star density get's too low put in G, too.

The chance (over all systems) to find an ELW is approx. 0.44 %. With the help of science, the rate to find ELW's can be tripled.

Edit: Error in the ordinate label of the second and third image. It is of course < Chance to find an ELW […] >.

With the exception of M type main stars had no masscode < A > system ELW's (and just seven ELW's were found in M-a systems). But as one can see above am I just considering masscode < b > to < e >. This, lead to an error in the original analysis because for M type star systems all columns were shifted by one position.
It was discovered by CMDR marx, who independently checked the results, that something was odd. Below I present the correct graph for the probabilities to find an ELW in a system with a given star type and mass code:

The only changes take place for M type star systems. These have in fact a very low probability to contain an ELW. Contrary to what I said above: Don't go there!
All other probabiliites are discussed above and the statements are still valid.

 

[marx]

Cool, thanks! A few days ago, I wanted to add the total main star count to the "Stars" sheet, to get some better probabilities in there, but I realised I don't have the May 30 data for that anymore. It was a great idea for you to do this after the FSS too, since the stars are auto-scanned now. (I recall something like thirty thousand ELWs in the total EDSM data which didn't have the main star scanned. Oof.)

Quite curious to see the A probability being slightly higher than the F. Neutron stars are a bigger surprise though: normally, I'd say because they are so well explored, and the FSS likely discouraged people from visiting M main stars... but this is adjusted for that.

The neutron main stars are also better than M dwarfs because they often contain other stars, around which the ELW can form and get heated by the NS too. This was another reason why I broke down the system categories: for example, there were 8,262 ELWs in systems with a main Neutron Star, but only 2,370 around a solo NS, and only 3,089 orbit the NS directly. Quite different from what you get with the AFGKM main sequence stars.
But yeah, since we can only filter on the galaxy map for the main stars, that's still the best bet.


Oh, and "lower total numbers (of course)": for a good while, this wasn't self-evident, as before the FSS, uploaded ELWs would often outnumber AWs! The reason for this was that people didn't scan the latter, most likely because they didn't recognize them. Plus it took more effort to scan them than it does now. Hm, now that I think about it, it would be interesting to see how many AWs are scanned but not mapped, versus how many ELWs are scanned but not mapped.

 

[marx]

WHOA! We have a clear winner: If you want to find ELW's, go to systems with an M type main star (red dwarf that is) and masscode < e >. That came as a total surprise, especially after seeing the absolute numbers above. It may be an artifact though, since the numbers are 43 ELWs found in 430 M-e systems. on the other hand, the numbers are high enough that I would risk a bet.

With the usual probabilities being as they are, I'd say that 430 systems is far too small a sample size. Although it could be that the chances there really are much better, it's just that this combination is so rare. But now I wonder about these M main star and mass code E systems, because I don't remember ever coming across one. Did you include the giant stars there? Edit: never mind, I realised that there aren't nearly enough M giants submitted to make a difference.

 

[schlowi123]

I agree with you and this is the reason why I estimate my chance to win this bet to be below 50 %.

On the other hand, for F / G / K stars we have of masscode < e > systems: 793 / 1161 / 593 and found there 5 / 13 / 9 ELW's respectively.
For M-e systems the quantity of total systems is approx. the same, but the number of ELW's found there 5 to ten times as high. It could still be a statistical fluke, but I'm confident enough to bet that in a year's time M-e systems will still have the best chance to find an ELW. And this even if all the M-e systems found up to 2019-06-18 are taken out of the analysis. If anybody takes this bet I offer to pay for a one paintjob (or paintjob bundle) if I lose.

Edit: This is an honest bet. Is there a possibility to make sure that I will pay up if I lose? Except that I'm an honest guy of course. I mean a third party that takes my money for the paintjob or a deposi for this new in-game currency or sth. like this.

Edit 2: I retract this bet since it was based on false information (see original post for details).

 

[Sapyx]

Surely any mass code "e", star class M, would be a red giant or red supergiant, rather than a red dwarf? Unless red dwarfs turn up in mass code e in the Cross of Suppression?

The only mass code d class M stars I have seen in my Survey have been Red Giants; I would assume this were true for mass code e as well.

 

[marx]

For M-e systems the quantity of total systems is approx. the same, but the number of ELW's found there 5 to ten times as high. It could still be a statistical fluke, but I'm confident enough to bet that in a year's time M-e systems will still have the best chance to find an ELW. And this even if all the M-e systems found up to 2019-06-18 are taken out of the analysis.

Actually, could you output the list of such M-e systems, both the total and the ones which have ELWs? I'd like to take a look through them, see what it might be about.

 

[Col Frost]

Thought you might be interested in this

 

[marx]

Actually, I've quickly run through Orvidius's data, and noticed that there are only 13 ELWs in mass code E, M dwarf main star systems. Seven of which are giants. Are you sure that your count was accurate? Could you output the ones it gave you?
For the record, the thirteen are:

Spoiler

Dryaa Blou PX-U e2-632
Dryoi Pri WE-R e4-6167
Eachaiv TO-Z e2844
Eulail CA-A e558
Kyloaqs EG-Y e2859
Myumbe XO-Z e1202
Phroi Pruae GA-A e9093
Cyoidai QY-A e2
Dryau Phyloea MI-B e3
Eembaisk WO-Z e5152
Plaa Eurk BL-P e5-4
Plaa Eurk BL-P e5-8
Shrogaei SI-B e194

When it comes to the giants, all the ELWs are orbiting secondary stars in the system. When it comes to the dwarfs, the systems all appear to be in the suppression cross. That doesn't sound surprising, but it's still pretty interesting.

 

[schlowi123]

I have the data on a different laptop. I'll get back to that.

It is of course possible, that I did a mistake in the code. If that is the case I will run the analysis again.

 

[Orvidius]

It's also possible I've screwed something up too, but the planet lists are usually fairly straightforward.

 

[schlowi123]

It was a copy and paste error . The number for M type main stars (but JUST those) are very different!

I've added this information at the end of the original post and made a note in the text that these specific results are wrong. I left it in there though, for the sake of discussion. I also don't like hiding mistakes … feels dishonest … and that would be the anti-thesis to the scientific method.

I thank CMDR marx for the independent analysis which in the end led to the discovery of this mistake.

I've also retracted the bet since it was based on false information.

I apologize for not checking my results good enough prior publication. … … … But it would have been nice if we would have found the ingame Genesis device.

 

[marx]

Kudos to being transparent about that. It's a lot more than many people can say, after all.

It's a pity that it was an error though. It would have been interesting to see something significant and new that was so far overlooked.

 

[schlowi123]

Thank you for the compliment. And it would have been indeed interesting to find something like that. Originally seeing these numbers didn't feel wrong. Since the overall number of M-e systems was so small I could totally imagine that FDev (or a rogue programmer) put a special something into the code.