Official Smogon University Usage Statistics Discussion Thread, mk.2

Status
Not open for further replies.
The Immortal, as far as I know, there is nothing in the logs that stores which Pokemon get selected. But the "real" usage corresponds to the number of times that Pokemon actually appeared in battle.
 

aVocado

I mean, where the fuck should I really even start?
is a Site Staff Alumnusis a Forum Moderator Alumnusis a Community Contributor Alumnusis a Tiering Contributor Alumnusis a Top Contributor Alumnus
In the most recent stats, Sylveon has no "Checks & Counters".. any reason for this?

edit: i'm talking about the 1825 OU ones.
 
Arikado,

The weighted stats tend to do that because they have a small sample size. Lower weighted stats should have some.
indeed, this is addressed by Entry 10 in the FAQ.

  • What's the best way to make use of the moveset stats?
    • If you're trying to figure out what's good in a tier (in terms of movesets), 1760/1825 is probably the way to go, since that tells you what the very top players use on their Pokemon.
    • If you want to determine what the likelihood is that your opponent's Pokemon carries X move or Y item, consult the moveset stats closest to your own Glicko R rating.
    • If you're having trouble dealing with a certain Pokemon and are looking for checks/counters, consult the 1500 (or even possibly the 0) stats: the lack of "1337"ness is vastly preferred to the sheer lack of data you encounter when you get that high.
 

Arcticblast

the real n1n1
is a Forum Moderatoris a Battle Simulator Driveris a Smogon Social Media Contributor Alumnusis a Super Moderator Alumnusis a Community Contributor Alumnusis a Tiering Contributor Alumnusis a Battle Simulator Moderator Alumnusis a Past SPL Champion
Moderator
Do you think it would be possible in future months to combine the usage of type-specific boosting items? There's no difference between them unless you're using Arceus (which is understandable, but Arceus types are listed separately) or Fling (which is a VERY rare move). Combining the items would, if I'm interpreting the script right, allow us to see more information on item usage than we already do. Granted I don't really know the technical details of how the stats list items or if it's even possible, but it's an idea.

<capefeather> !usage1337 talonflame items
<TIBot> Choice Band 56.147% | Leftovers 15.940% | Life Orb 11.521% | Sharp Beak 8.073% | Sky Plate 3.320% | Other 4.999%
 
Is there a reason why the Stats/2014-07/mega files for OU and RU-Beta are empty? They are also listed to have filesize of 0.
 
Seismitoad, no--there is no plan to support hackmons/BH. I also think there are issues with OU Theorymons.

Gen IV is giving me some issues as well, and *that* I may try to address, but yeah, no love for the hackmons.
 
I don't know if it was brought to your attention when an above user pointed out certain empty files, but several in the moveset directory for this month are empty, including ou-1500, which is one I can't see being empty due to lack of data.
 
Shayhok, outheorymon apparently falls into the same category of hackmons (breaks my scripts, so I'm not gonna bother). OU-1500 and randbats-1500 were two jobs I had to redo, and I simply hadn't gotten around to uploading the data before today (OU is up, randbats I still need to rerun).

The other missing data appear to be previous-generation metas, so I'm guessing the issue really was: not enough data.
 
In their current state, the RU changes since last month stats aren't too useful since June's stats were almost entirely made up of the suspect stats while July's were almost entirely made up of non-suspect stats, leading to the changes section for both the RU and RU suspect test stats to be comparing almost a whole month's worth of data to only a few days worth. Would it be possible to have stats comparing June's RU suspect stats to July's combined RU stats?
 
Just a heads-up: the weighted 1v1 stats for August use the same listing order as the unweighted stats, despite the percentages being vastly different (resulting in, for example, Whimsicott at 34th place with 1.22547%, and then Sylveon at 35th place with 13.86890%).
 
Just a heads-up: the weighted 1v1 stats for August use the same listing order as the unweighted stats, despite the percentages being vastly different (resulting in, for example, Whimsicott at 34th place with 1.22547%, and then Sylveon at 35th place with 13.86890%).
I think this was a deliberate choice (they're sorted by Real, meaning the number of times the Pokemon *actually* appears in a battle). But Real appears to be unweighted, so that's wrong, at least.
 
Antar I hope you can clarify this for me.

I want to analyze teammates data by calculating their P(teammate | pokemon) values. My code will be limited to reading the JSON files only (not going to write a parser for the moveset TXTs), so that's all I'll be working from. Considering what you said before:
QxC4eva, the number in the JSON is the number in the TXT times Aegislash's usage:

.04865 x 55176 = 2684

(55176 is the 1695 usage count for Aegislash, which is most easily calculable by taking the sum of the Abilities values)
... suggests that
txtValue * sum{PokemonAbilities} = jsonValue
=> txtValue = jsonValue / sum{PokemonAbilities}


P(X|Y) would be that value in the TXT plus the usage percentage.
P(teammate | pokemon) = txtValue + teammateUsage%
= jsonValue / sum{PokemonAbilities} + teammateUsage%


Is this correct?
(If you have any documentation etc. that would be helpful)
 
Considering what you said before:

... suggests that
txtValue * sum{PokemonAbilities} = jsonValue
=> txtValue = jsonValue / sum{PokemonAbilities}
I *believe* that's correct, but a great way to check is to try three or four examples and verify that you get the correct numbers by checking the txt (don't blame you for not wanting to parse--that's why I made the json in the first place).

P(teammate | pokemon) = txtValue + teammateUsage%
= jsonValue / sum{PokemonAbilities} + teammateUsage%


Is this correct?
If the first part pans out, then the second should as well.

(If you have any documentation etc. that would be helpful)
Yes. Documentation would be a very good idea. But--real talk--it's not coming any time soon.
 
Ugh I was hoping you say I was wrong lol, the results I got are incorrect. But I think I figured out what you did..

Also, what's been happening with this?
Well *I* know how many were KOs vs. switch-outs. It was just an oversight that I didn't provide it in the json. I'll try to remember to rectify that before the next update (might as well provide all the matchup data).
 
You know what would be great? If Hackmons and Balanced Hackmons tier wasn't ignored since February. I'm very interested in seeing how things are in that tier, but it's always dismissed. Is there a reason for that?

EDIT: I'm referring to this btw; It's absent here unlike every other active tier. http://sim.smogon.com:8080/Stats/
 
You know what would be great? If Hackmons and Balanced Hackmons tier wasn't ignored since February. I'm very interested in seeing how things are in that tier, but it's always dismissed. Is there a reason for that?

EDIT: I'm referring to this btw; It's absent here unlike every other active tier. http://sim.smogon.com:8080/Stats/
Hackmons breaks my scripts. Specifically, IIRC, it's the fact that you can have megas on your team (without needing to mega evolve into them). So I'm sorry, but there will be no hackmons stats any time soon.
 
Hackmons breaks my scripts. Specifically, IIRC, it's the fact that you can have megas on your team (without needing to mega evolve into them). So I'm sorry, but there will be no hackmons stats any time soon.
Hackmons breaks my scripts. Specifically, IIRC, it's the fact that you can have megas on your team (without needing to mega evolve into them). So I'm sorry, but there will be no hackmons stats any time soon.
Yeah that's right. In Hackmons you can have mega's in your team. I also remembered that it took a lot of info even down to 0.500% of move usage for a single pokemon. If it breaks scrips (as you say) now, then how did you bypass it before? Pokemon like Diancie and Mega Latios has been introduced since the last time Hackmons/balanced Hackmons was listed. If mega's is the issue, shouldn't capmons statistics interrupt as well? I think everything should be deserving for a solution.
 
Status
Not open for further replies.

Users Who Are Viewing This Thread (Users: 1, Guests: 0)

Top