Czech names generated/used by the game

[geckon]

I just finished selection of Czech first names - I have 472 of them. Although I may decide to remove some of them during the frequency assigning.
I keep there a few Polish, Germanic, Hungarian names and also some probably popular names that are not Czech (like the mentioned Nikolas, Sebastian or Leonard) and variants of Czech and foreign names (like Dominik - Dominic, Nikolas - Nicolas, Kvido - Quido etc.).
If you have any advices or comments of the bit I just wrote please let me know. Otherwise please wait until I complete and upload it so you can check it out.

PS: I found nice source for surnames and will do them as the next thing.

[grindline]

Following on from this I was thinking about the English name sets. Given that a statistically noteworthy amount of the English population are ethnically Asian, I'm thinking specifically about the Pakistani Muslim community, should there be the most common names among that demographic?

Pro:

It would be more realistic to have an English regen called Mohammed Hussain

Cons:

How do you guarantee the ethnic first and surnames match up so that you don't get super rare (in real life) cross ethnicity names such as Jason Mohammed?

In the event that facegen appears in FHM would there be potential reality issues with having an obviously ethnic name being assigned to a Caucasian face?

[geckon]

Czech first names done. I have 455 of them. The file is attached to this post so you can look at it. I am open to your suggestions.

The names come from Aktuální a podrobná statistika jmen (Czech) and their frequencies are based on the real data but tweaked according to JeffR's advices in this thread.

Don't you have a piece of software for generating players names? I could use something like it to test it. I'd like to generate bunch of names and check it out to avoid the possibility that the file is compiled wrong way. In that case I could try and repair it.

Originally Posted by grindline

Following on from this I was thinking about the English name sets. Given that a statistically noteworthy amount of the English population are ethnically Asian, I'm thinking specifically about the Pakistani Muslim community, should there be the most common names among that demographic?

Pro:

It would be more realistic to have an English regen called Mohammed Hussain

Cons:

How do you guarantee the ethnic first and surnames match up so that you don't get super rare (in real life) cross ethnicity names such as Jason Mohammed?

In the event that facegen appears in FHM would there be potential reality issues with having an obviously ethnic name being assigned to a Caucasian face?

While it is true, it's not uncommon to find English people with Asian names. Finding asian hockey players in England is a different matter!

[geckon]

OK now I have finished the last names as well. Here they are (all 10.000 of them).

The file is based on this source: Četnost českých příjmení (Czech).

There are some (not many) Vietnamese and foreign (e.g. Polish or German) surnames which I think is completely OK because such surnames are used in Czech republic. Vietnamese surnames are used only by Vietnamese people but their children will play hockey as well.
I have modified frequencies in a few ways...
- The most common names' priorities were lowered.
- The Vietnamese and some other "less-Czech" surnames were lowered notably.
- The surnames of famous hockey players from the real world were somewhat lowered too.
- Other smaller changes.

Now it would be nice to generate some names to test whether the set is done right. And of course you are welcome to go through the files and point out possible mistakes or ideas for improving the set.

[YZG]

Originally Posted by geckon

That would be really cool but I am not sure if it wouldn't also be difficult to ensure that some nationalities can generate a player with Vietnamese/Inuit/Tatar/etc name. JeffR already stated that Czech player has certain chance to get Slovak name and I don't know how far they go in that but he also said something about bloating the world config file (when I suggested something similar).
Also I don't feel I'd be able to create let's say Vietnamese or Romani name file of high quality.

Yes... might be best to wait to know precisely how this is going to work before going too far.

Originally Posted by geckon

If you are interested to help and want to know what has been done and what hasn't you can look at the file attached to this JeffR's post:

Thanks, I had missed that post. From what I can see in there, I can contribute right away the Slovenian set (424 forenames + 910 surnames; credits go to Bladerunner from TBL who compiled these for me a while ago). From the same source, I have a small (368 + 517) Serbian set available. I can look to expand it a little.

I noticed a few sets are missing. Given what's already planned in there, IMO there should also be Bulgarian, Turkish, Greek, Mongolian and Armenian ones. There should also be something from South Africa. Belgian should also be split into Belgian Dutch and Belgian French, similar to Switzerland. I reckon that adds a fair amount of work for it all to be ready for release day, though... Perhaps as a bonus if time allows, and/or a post-release expansion?

I have a small Bulgarian set ready that could do it (240 + 517) or be expanded as needed. The others wouldn't be very long and difficult to build, especially since I think they might not need to be as thorough as the bigger nations, at least for the time being.

Other than that, I currently have (South) Korean in progress. It should be ready this weekend; chances are, I can clear a second one by Monday as well.

- YZG

[JeffR]

Originally Posted by grindline

How do you guarantee the ethnic first and surnames match up so that you don't get super rare (in real life) cross ethnicity names such as Jason Mohammed?

We can't, unfortunately. So, for that reason, keep it as ethnically English (or Scottish, or Irish) as possible. If we want, we can reflect immigrant populations by giving a very small percentage of English players names from, say, the West African, Arab or Punjabi lists.

In the event that facegen appears in FHM would there be potential reality issues with having an obviously ethnic name being assigned to a Caucasian face?

Yup. It's going to be an unavoidable problem, at least for now, in places like the US and Canadian name lists that have to reflect ethnic populations that tend to have anglicized first names - so the names will come out right, like Paul Kariya or Richard Park, but the facegen faces won't match.

[JeffR]

Originally Posted by geckon

Don't you have a piece of software for generating players names? I could use something like it to test it. I'd like to generate bunch of names and check it out to avoid the possibility that the file is compiled wrong way. In that case I could try and repair it.

No, this looks good - I can't test it yet in the game because it's going to require a bunch of changes to the world.xml file to switch over from the temporary OOTP files we're using, but I don't see any problems with it. And 450 names is a big improvement over the 130 or so the original file had, nice work.

[JeffR]

Originally Posted by geckon

Now it would be nice to generate some names to test whether the set is done right. And of course you are welcome to go through the files and point out possible mistakes or ideas for improving the set.

Here's 20 random names (just done in a spreadsheet, not the game, but using the actual game formulas). How do they look?

Ladislav Rut
Vladislav Havlíček
Jindřich Hrubec
Ladislav Nosek
Petr Fojtík
Vojtěch Heřmánek
Antonín Štěpaník
Aleš Rybka
Petr Dočkal
Patrik Potěšil
Ladislav Řehoř
Oldřich Vaca
David Sitár
René Petráš
Karel Křížek
Martin Novák
Marián Schuster
Miloš Hanek
Antonín Nesvadba
Karel Hofman

[grindline]

Just out of interest how many names do we expect to find in a set? The English first names are currently at 250, which seems a bit low but covers pretty much all of the names that have a frequency of 0.25% or greater in real life.

I guess I can add a few more rare ones to go down to those with a frequency of 0.1% if that is in keeping with the other name sets?

[archibalduk]

Originally Posted by grindline

Just out of interest how many names do we expect to find in a set? The English first names are currently at 250, which seems a bit low but covers pretty much all of the names that have a frequency of 0.25% or greater in real life.

I guess I can add a few more rare ones to go down to those with a frequency of 0.1% if that is in keeping with the other name sets?

What data source are you using for the frequencies?

[grindline]

Originally Posted by archibalduk

What data source are you using for the frequencies?

The table 6 from the 2011 census. Jack is c.7500 out of a total of c.340,000 therefore about 2%.

[archibalduk]

Originally Posted by grindline

The table 6 from the 2011 census. Jack is c.7500 out of a total of c.340,000 therefore about 2%.

By my calculations (and they seem to match yours - I have Jack at 1.98%), the top 250 takes us down to Kane which has a frequency of 0.06%. I only see 77 names at 0.25% and higher. I don't see 250 names with a frequency of at least 0.25%.

If we're using Table 6, I'd suggest widening the net somewhat. For example, the name "Fred" ought to be included IMO - he's at #576 (0.02%). What do you think of going down to about #1158 where we have Roy (i.e. those with a count of at least 22)? Once the non-English/Welsh sounding names are taken out, I think we'd be left with just under 1000 names - which is the equivalent to just under two distinct names per English/Welsh player in the game DB.

Off-topic a little, but I can't believe so few babies were called the likes of "Roy" in 2011. But then I guess that's a sign of the times and/or me getting old!

[geckon]

Originally Posted by JeffR

Here's 20 random names (just done in a spreadsheet, not the game, but using the actual game formulas). How do they look?

(...)
Petr Fojtík
(...)

I think they look very good. Of course for precise testing it would be needed to watch generated names for some time (as new players come) so it would show potential problems in a long run. But for now I think it's OK. I think it can be upgraded anytime later if such problems show right in the game.
BTW Petr Fojtík is my actual friend

Originally Posted by JeffR

Yup. It's going to be an unavoidable problem, at least for now, in places like the US and Canadian name lists that have to reflect ethnic populations that tend to have anglicized first names - so the names will come out right, like Paul Kariya or Richard Park, but the facegen faces won't match.

Does this post indicate that we can look forward to the facegen or is it meant like "if there will be the facegen it will suffer on this"?

[archibalduk]

Originally Posted by geckon

Does this post indicate that we can look forward to the facegen or is it meant like "if there will be the facegen it will suffer on this"?

I think Sebastian posted somewhere here a while back that they would look into FaceGen, but it's not yet certain whether or not it will make it in to the game.

[geckon]

Originally Posted by archibalduk

I think Sebastian posted somewhere here a while back that they would look into FaceGen, but it's not yet certain whether or not it will make it in to the game.

Yes, I remember something like that too and I want to know if it has been decided already.

[geckon]

Could a moderator edit this thread's title to reflect the fact that not only Czech names are the subject of discussion here? I would do it myself but haven't found out how and I'm not sure if I have the needed permissions.

[grindline]

Originally Posted by archibalduk

By my calculations (and they seem to match yours - I have Jack at 1.98%), the top 250 takes us down to Kane which has a frequency of 0.06%. I only see 77 names at 0.25% and higher. I don't see 250 names with a frequency of at least 0.25%.

That is what happens when I do mental arithmetic with a child shouting at me. I was calculating a tenth of a percent which was c.300 and then halved it to get c.150 but when writing the frequency I accidentally doubled instead of halved so I meant to write 0.05% not 0.2%

Duh!

Oh and as for expanding the database, my thoughts were to do all the English names with 50 or more occurances then add any others of note below that point, similar to what you suggested.

[archibalduk]

I've completed the Irish and Scottish names and have sent them to Grindline for double-checking. Once he's okay'ed them, I'll put them into a text file using the correct format.

I have frequency figures from the 2011 census for first names for both countries. There aren't any comprehensive frequency figures for surnames, so they'll each need to be set to '1' like the English surnames.

[YZG]

Are Asian names in-game intended to use the Given name /Surname format or the Surname / Given name one?

I am joining two csv sets here: first one is a small Bulgarian set (239 + 552), if you're interested to add it. It was built from a couple of lists of people and names are weighted according to their occurence in said lists.

The second is a Korean one (607 + 182). I checked to reuse the one that comes with OOTP, but found out that, like the Russian set, it includes both male and female names. Moreover, the names are transliterated according to several romanization systems. So, I rebuilt the set anew. It features 182 of the ±250 known Korean surnames and they are weighted according to real-life proportions. Given names have been compiled from various lists of people. A few have been made more common according to frequency of occurence in lists and according to data on the most common given names to babies for the past few decades. I also applied the Revised Romanization of Korean to all gathered names to ensure names are uniform. I see some names are slightly different in North Korea, but it might not be worth the hassle for now.

I see the default OOTP Chinese names are a real mess as well. Seems like there's a mix of Chinese and Taiwanese names in there. I'll build a fresh new one.

- YZG