Overall Rating vs Usage

[Paperfairy]

I have a huge math project due in the near future. The project requires us to carry out a mathematical investigation that uses math that is covered in our course. (I am in a pretty lame Math class, so, the Math is pretty lame.)

I was curious as to whether or not Smogoners would be willing to proofread and help me develop a better conclusion, and anything else you'd like to add or suggest.

Thanks, enjoy~!

Ver 1 (OU only) http://h1.ripway.com/paperfairy/IA.pdf
Ver 2 (All but Ubers) http://h1.ripway.com/paperfairy/493IA.pdf

EDIT: For those of you that speak IB, this is my Mathematical Studies SL Internal Assessment.

 

[Caelum]

Well, the document link is broken.

This forum isn't for correcting your homework; however, if it's something that could promote relevant discussion this thread might be okay (e.g. a math oriented Pokemon thread like X-Act's) - but I can't tell since the link is broken. Please fix it up.

 

[Paperfairy]

I suppose you've a point - I thought the discussion _could_ be relevant... I suppose you can be the judge of that.

http://h1.ripway.com/paperfairy/IA.pdf

 

[Caelum]

OK, let's see if we can meet middle ground here.

Let's try to make the discussion of what makes certain Pokemon who have excellent stats see relatively limited usage whereas Pokemon with lower rated stats see substantial usage. I can conjure up a bunch of reasons like ability, move pool, typing etc; but maybe this could be a dissection of specific Pokemon rather than just generalized reasons.

And if someone wants to correct something in your project they can do that.

Nice job though, I just skimmed it so I might have missed something but it seemed mathematically solid.

 

[Paperfairy]

Sounds good - thanks.

While I am not shocked at Scizor being at the top of usage (because BulletPunch just rapes), Cresselia being 44th was genuinely surprising. I also question the legitimacy of all of the Rotom formes being in this study, considering that most of them saw usage in ranks of 60~ in the Starndard ladder. Thoughts on that?

 

[Eraddd]

Is this your Math SL portfolio?

I'd say this is a huge risk because IB markers sometimes known to be conservative, as in, they might not look too highly upon a subject that is to them a kid's subject.

Other than that, your portfolio does look fine.

Edit: Oh and you might want to elaborate more on the variables that were not discussed in your portfolio as in moveset, abilities etc, becuase if the markers are poke-dumb, they won't understand the variables, and might dock some points.

 

[Paperfairy]

Is this your Math SL portfolio?

I'd say this is a huge risk because IB markers sometimes known to be conservative, as in, they might not look too highly upon a subject that is to them a kid's subject.

Other than that, your portfolio does look fine.

Edit: Oh and you might want to elaborate more on the variables that were not discussed in your portfolio as in moveset, abilities etc, becuase if the markers are poke-dumb, they won't understand the variables, and might dock some points.

I would agree, except we had a student from our school get a 19/20. Her project: Does the number of players in a game of Candy Land impact the length of a game?

I agree with the second point... but how can I elaborate without writing a primer on competitive battling?

 

[Tangerine]

I think your conclusion was weak particularly because you were ONLY using OU. If you used all of the other Pokemon, your r should have definitely been stronger, and you probably would have seen a much stronger shape to work with (I personally don't think it'd be linear but closer to an exponential curve). Adding the overall usage stats for the rest of the critters, and running the same analysis would lead to a much stronger conclusion. If this was to be repeated then with the UU tier, etc etc, I think you can have a much better read on how people pick their Pokemon - because people DO look at stats when they choose Pokemon.

I think you should just emphasize that base stats are one of the reasons why people choose their Pokemon, and that'll solve your problem of having to explain ability/typing/etc... and by adding the UU/NU critters, you can take critters such as Regigigas as an "example" since he's obviously going to be an outlier if we're going by overall rating only, which would clarify your case a bit more.

 

[Paperfairy]

So use all of the Pokemon, but only look at the Standard ladder? That's more work, but I'm not against that if it produces a stronger result. Thanks for the input.

 

[tubbablubba]

i just got my Ib diploma earlier this year and the math looks fine here, the only problem here is trying to make it as non-player friendly as possible. I think the OU test sample is fine but it needs to be clearer in its descriptions of common thought in the pokemon community that others dont comprehend so easily.

 

[Mazaki Anzu]

I dont know much about matchs and cant open your file so I'm not quite sure what's going on at all, sorry about that.

But I would like to give a suggestion regarding movesets.

STAB should be take in count of course

Power (70+, 80+, 90+ and 100+) also, type coverage (like Drapion with EQ to hit fire, just an example) also and most important offensive stat to use it. Effects like recoil or charging turn/reload turn should give a negative bonus.

Like for example Entei's fire fang got STAB and use it's best offensive stat but does not have the power to be counted, so it won't give a good bonus to its final punctuation as ou, uu or whatever.

Traits also should be take in count, no effect in battle, the need of turns to work (Slow start, truant) or an ability that gives a pokemon recovery or a boost in it's stat (flash fire, huge power, marvel scale)

 

[Paperfairy]

I'll work on that. And congrats on the IB diploma.

I also just looked at the July stats - there are only 138 on usage stats, ending at 3838... so I cant give the remaining Pokemon a zero. Looks like I may have to restrict myself to a tier.

EDIT: Anzu, tis a standard .pdf file.. do you have a PDF viewer installed? Movesets were not included because you cannot quantify that data.
EDIT2: Turns out 480 Pokemon appear... my page hadn't finished loading.

 

[iss]

um, on the second page:

Statement of Tas

shouldn't it be task?

 

[Paperfairy]

Yeah. Thanks. Changed it. Also turns out that Scizor's base speed was all wrong.

 

[lordkira]

I would agree, except we had a student from our school get a 19/20. Her project: Does the number of players in a game of Candy Land impact the length of a game?

I agree with the second point... but how can I elaborate without writing a primer on competitive battling?

It's fine I got a very high score in IB TOK when my friends and I did our project about how we can get ripped with six pack in 2 weeks by working out and eating a can of tuna and some salad with some egg yolks everyday. It was ridiculous but in my honest opinion it just has to be well thought out...and uniqueness definitely helps.

 

[Eraddd]

I would agree, except we had a student from our school get a 19/20. Her project: Does the number of players in a game of Candy Land impact the length of a game?

I agree with the second point... but how can I elaborate without writing a primer on competitive battling?

lol really? I know a guy who tried to do theirs on a video game, can't remember, but was told by the IB coordinator to not to do it.

When you compared Scizor to Cresselia, maybe you can put in note the moveset? Like mention bullet punch, with it's bonus, and the ability to put a dent in most of the commonly used pokemon, and cresselia, the supposed wall, has a crappy recovery move, and is specially hit by the most common types in the metagame (bug and dark).

Actually, now that I think of it, our IB coordinator is pretty darn strict and conservative, so I guess that's one factor everyone tries to stick to used topics.

 

[Paperfairy]

Good idea with the Scizor vs Cresslia. And I think I've decided to keep the study at OU, but mention in the conclusion what may have happened if I had done all 480.

 

[Burn Jack]

I don't think it's fair to conclude that "there is barely a relationship between the overall rating of a Pokémon and its actual usage in battle", given that you have excluded a whole raft of Pokémon on the grounds that they are not used enough.

Also, I don't think the final sentence, "Only 53 Pokémon..." says what you mean it too.

 

[Anachronism]

I think that your idea differs with the candyland one in that more people have played candyland. Very few people have even heard of competetive pokemon let alone usage stats etc. There is just too much explanation that goes into it beforehand. I think you should do the IB IA on something familiar to more people or at least easier to explain so you can spend more timeon the math.

Are you in SL 1 or 2? They didn't let us choose our topic in my SL 1 class. We had to create a function of best fit for the graphs of data on crows dropping nuts from various sizes lol. The graphs were height of drop vs. how many drops it took to crack the nut on average. You are lucky you get to choose what you want. Choose wisely (as in not this).

Even though you did a great job on the math, I worry that this won't get a good score. But then again, getting an A on this doesn't matter if you get A's on the tests (like me :) so don't stress too much.

Also, I think you final statement needs some revision. Other than that, you have a very solid paper. I hope that the subject doesn't cause point deductions.

This paper is also rather interesting. The distribution is quite abnormal.

 

[petrie911]

Firstly, despite how the statistics are gathered, Smogon considers all 5 Rotom formes to be 1 Pokemon. Thus, you should add their usages together and consider them to be the single Pokemon Rotom-A

Secondly, OU is a poor cutoff point to use, as the idea that only OU Pokemon will be found on OU teams is simply wrong. There are many quite good UUs that can find their way onto serious OU teams, such as Dugtrio, Yanmega, Roserade, Donphan, Milotic, Arcanine, and others. A better cutoff point would be X-Act's metagame snapshot, ie, the set of pokemon such that, when you add up their percentages, it totals to over 500%. That means that a "standard" team consists of choosing 6 from that set. BTW, assuming all Rotom formes count as one, that puts the cutoff at #61 (Roserade). Going a bit further along to 2% usage isn't too bad of a place either, making the cutoff #66 (Arcanine).

Lastly, the several non-stat-related variables should be elaborated on, so those unfamiliar with the game would understand their effect. Perhaps you could even suggest a modification to X-Act's rating system that would account for these things and improve the correlation. Of course, I'm not sure if this is the sort of thing IB would look for, so this may not be wholly necessary.

 

[Paperfairy]

Firstly, despite how the statistics are gathered, Smogon considers all 5 Rotom formes to be 1 Pokemon. Thus, you should add their usages together and consider them to be the single Pokemon Rotom-A

The official tier list says otherwise.

Secondly, OU is a poor cutoff point to use, as the idea that only OU Pokemon will be found on OU teams is simply wrong. There are many quite good UUs that can find their way onto serious OU teams, such as Dugtrio, Yanmega, Roserade, Donphan, Milotic, Arcanine, and others. A better cutoff point would be X-Act's metagame snapshot, ie, the set of pokemon such that, when you add up their percentages, it totals to over 500%. That means that a "standard" team consists of choosing 6 from that set. BTW, assuming all Rotom formes count as one, that puts the cutoff at #61 (Roserade). Going a bit further along to 2% usage isn't too bad of a place either, making the cutoff #66 (Arcanine).

That cutoff point was actually my original idea, but in the end, I decided to do all but ubers. Link in the OP.


Lastly, the several non-stat-related variables should be elaborated on, so those unfamiliar with the game would understand their effect. Perhaps you could even suggest a modification to X-Act's rating system that would account for these things and improve the correlation. Of course, I'm not sure if this is the sort of thing IB would look for, so this may not be wholly necessary.

I did some more as you suggested, but I really don't know if I have gone overboard.

I think that your idea differs with the candyland one in that more people have played candyland. Very few people have even heard of competetive pokemon let alone usage stats etc. There is just too much explanation that goes into it beforehand. I think you should do the IB IA on something familiar to more people or at least easier to explain so you can spend more timeon the math.

I would agree if my math teacher had not already approved it.

Are you in SL 1 or 2? They didn't let us choose our topic in my SL 1 class. We had to create a function of best fit for the graphs of data on crows dropping nuts from various sizes lol. The graphs were height of drop vs. how many drops it took to crack the nut on average. You are lucky you get to choose what you want. Choose wisely (as in not this).

IB teachers are notorious for doing that - lying to their students and forcing topics, but IB says you can always choose topics for IA/EA.

Even though you did a great job on the math, I worry that this won't get a good score. But then again, getting an A on this doesn't matter if you get A's on the tests (like me :) so don't stress too much.

I got a hold of last year's exam and would have scored a 7. I'm confident.

Also, I think you final statement needs some revision. Other than that, you have a very solid paper. I hope that the subject doesn't cause point deductions.

Please read revised copy, as I have elaborated on that.

This paper is also rather interesting. The distribution is quite abnormal.
Indeed!

I don't think it's fair to conclude that "there is barely a relationship between the overall rating of a Pokémon and its actual usage in battle", given that you have excluded a whole raft of Pokémon on the grounds that they are not used enough.

Agreed, so I did the project over.

Also, I don't think the final sentence, "Only 53 Pokémon..." says what you mean it too.

Removed, but you were right.

Anyways, I went ahead and redid the project with everybody but Ubers. The results are still rather shocking - that stats play a significantly less important role than I originally thought. Moves and abilities really do make the difference.

Ver 2.0: http://h1.ripway.com/paperfairy/493IA.pdf

 

[Anachronism]

I think that doing all pokes is a litttle excessive. Now I think you may have obscured your observations with the sheer mass of your data. Even doing the top 100 or 200 would have been more than adequate, but including all of them?! that is just excessive. I would figure out what your interquartile range is and remove the lower outliers that way. The thing though that the average would probably be low because of all the pokes with low usage, meaning stuff that isn't remotely competetive would be included. You should see what happens though when you create a box-and-whisker plot of the usage data.

It is nice to hear that your teacher already approved, because that would be something I would not want to deal with when writing a paper of this magnitude.

I think that although the second draft is much better written than the first, you still need to adjust your scope to have an acceptable paper.

 

[BL Fighter]

Hey, just a heads up, Slaking is listed twice in the Overall ratings spreadsheet. other than that, I didnt see anything else.

 

[petrie911]

Firstly, despite how the statistics are gathered, Smogon considers all 5 Rotom formes to be 1 Pokemon. Thus, you should add their usages together and consider them to be the single Pokemon Rotom-A

The official tier list says otherwise.

For the sake of completeness, here is the entire OU list:

Aerodactyl
Alakazam
Azelf
Blissey
Breloom
Bronzong
Celebi
Cresselia
Dragonite
Dusknoir
Electivire
Empoleon
Flygon
Forretress
Gengar
Gliscor
Gyarados
Heatran
Heracross
Hippowdon
Infernape
Jirachi
Jolteon
Kingdra
Latias
Lucario
Machamp
Magnezone
Mamoswine
Metagross
Ninjask
Porygon-Z
Rhyperior
Rotom-A
Salamence
Scizor
Skarmory
Smeargle
Snorlax
Starmie
Suicune
Swampert
Tentacruel
Togekiss
Tyranitar
Umbreon
Vaporeon
Weavile
Zapdos

(49 Pokemon)

Unless I'm very much mistaken, that's Rotom-A on that list, not Rotom-H, -W, -F, -S, and -C. So, yeah, the official tier list agrees with me.

 

[Paperfairy]

Anachronism

That's an interesting idea. I submitted the draft to my teacher - I'll see what he thinks of your idea.

BL Fighter

Wow thanks. :D

petrie911

Oooh, okay. I was looking at the tier list on the website. I'll consider it.