Smash IV Data Collection Project Discussion

[Dr. James Rustles]

All good stuff, I think the one thing I'd like added is stage banned, it's VERY useful to know.

You can deduce what stages are banned by the ruleset check and the deviations list.

 

[ぱみゅ]

*Brainstorming*
Another thing it would be interesting to add to the sheets is Round, as in, have in consideration if it was Winners Round 2 or Losers Round 5, or GFs, so we could weight results more as the bracket advances.
With that, bigger tournaments will have more weight, and maybe we could also add some friendlies and MMs just for the sake of gathering data, even if the result doesn't really has an impact at the end.

 

[Overswarm]

you guys saw this right:

Has bracket round, stage bans, whole shebang.

 

[ぱみゅ]

pretty sure I missed that because the template was blank

A couple format/presentation changes, but overall looks pretty solid.

 

[El Duderino]

So just a quick observation, where do Melee and 64 fit into this research? It would make sense to take the entire smash community into account.

 

[Ulevo]

Well personally I don't see the need, at least as far as Smash Wii U is concerned. If someone wanted to do a similar project for Melee/Brawl/64, be my guest. Just go and post it in the respective forums. To be honest though, if Smash Wii U is good, nobody very few people are going to care about Brawl anymore, so that in of itself would seem like a wasted effort.

EDIT: Just realized you didn't mention Brawl. Watevs.

 

[El Duderino]

Oh... ok, I see. Saw these charts with Brawl specific characters and assumed people where attempting to gather this type of data on it first.

 

[Jack Kieser]

I approve of the concept, and think it will be plenty iterated upon before release. Also hope to see that full software package talked about earlier; many TO's at least have an internet connection at events, so it'd be nice to have a software package that can upload to the database directly.

The thing I'm still iffy on is the "no free tournaments" thing. Here's why. I get what OS is saying about "junk data" concerns, but I don't think it's actually junk data. First of all, if our concern is people not playing their absolute best all the time, why are we collecting data for anything less than Top 32 / 16 / 8? Anything less than that is below the top echelon of gameplay, anyway. If the only thing we're concerned about is absolute top play, then only collect absolute top data. If we want to collect the rest of a paid tournament, we should be fine collecting other non-optimal combat data, too. Sure, you could say that collecting some junk data doesn't necessitate collecting all junk data, but then again, we also know that far fewer low tier characters are played in paid-events. Low-tier events can be held for a fee, but they get less entrants.

I honestly think that the concern about players not doing their best is overstated, and by ignoring those players and events, we potentially lock ourselves out of data that, while less valuable than optimally collected top play data, is still useful to at least some of the community. If the database is built properly, it'd be trivial to separate the data when viewing it, so why not support the collection of it? If someone has a use for it, they can use it. And, it sends a good message to those players that we at least care enough to consider them.

 

[Zonderion]

I approve of the concept, and think it will be plenty iterated upon before release. Also hope to see that full software package talked about earlier; many TO's at least have an internet connection at events, so it'd be nice to have a software package that can upload to the database directly.

The thing I'm still iffy on is the "no free tournaments" thing. Here's why. I get what OS is saying about "junk data" concerns, but I don't think it's actually junk data. First of all, if our concern is people not playing their absolute best all the time, why are we collecting data for anything less than Top 32 / 16 / 8? Anything less than that is below the top echelon of gameplay, anyway. If the only thing we're concerned about is absolute top play, then only collect absolute top data. If we want to collect the rest of a paid tournament, we should be fine collecting other non-optimal combat data, too. Sure, you could say that collecting some junk data doesn't necessitate collecting all junk data, but then again, we also know that far fewer low tier characters are played in paid-events. Low-tier events can be held for a fee, but they get less entrants.

I honestly think that the concern about players not doing their best is overstated, and by ignoring those players and events, we potentially lock ourselves out of data that, while less valuable than optimally collected top play data, is still useful to at least some of the community. If the database is built properly, it'd be trivial to separate the data when viewing it, so why not support the collection of it? If someone has a use for it, they can use it. And, it sends a good message to those players that we at least care enough to consider them.

Well put Jack. But from the way things are going, it doesn't look like much of a software package for the TOs. It looks like its going to just be a spreadsheet that someone will have the privileges of doing all the grunt work. I stated earlier that I wish for it to be a system in the hands of the TOs to collect the data. But hey.

 

[Overswarm]

I approve of the concept, and think it will be plenty iterated upon before release. Also hope to see that full software package talked about earlier; many TO's at least have an internet connection at events, so it'd be nice to have a software package that can upload to the database directly.

The thing I'm still iffy on is the "no free tournaments" thing. Here's why. I get what OS is saying about "junk data" concerns, but I don't think it's actually junk data. First of all, if our concern is people not playing their absolute best all the time, why are we collecting data for anything less than Top 32 / 16 / 8? Anything less than that is below the top echelon of gameplay, anyway. If the only thing we're concerned about is absolute top play, then only collect absolute top data. If we want to collect the rest of a paid tournament, we should be fine collecting other non-optimal combat data, too. Sure, you could say that collecting some junk data doesn't necessitate collecting all junk data, but then again, we also know that far fewer low tier characters are played in paid-events. Low-tier events can be held for a fee, but they get less entrants.

I honestly think that the concern about players not doing their best is overstated, and by ignoring those players and events, we potentially lock ourselves out of data that, while less valuable than optimally collected top play data, is still useful to at least some of the community. If the database is built properly, it'd be trivial to separate the data when viewing it, so why not support the collection of it? If someone has a use for it, they can use it. And, it sends a good message to those players that we at least care enough to consider them.

All that data has to be manually entered by someone, so it's non-trivial adding extra data points. We want as few as possible while getting as much info as possible. Free tournaments not only need to be separated from tournaments, but have an asterisk next to them that requires MORE data to determine if its valid or not. I can play MK in tournament for Brawl, and I can play MK in a free-to-play tournament. In the F2P tournament there's nothing against me experimenting with other characters or tournaments or even just saying "I wonder if I can win with only B moves" (I have done this). It's basically junk data and tells us as much as a poll asking "who is your best character?"

Since we're collecting rounds and automatically player placements, we'd be able to filter easily to top 32, top 16, top 8, etc., to determine what trends are occurring. Do top tiers really help? We can see by filtering the data in this way. With free tournaments and smashfests we can't filter the data.

 

[Mormon Mammoth]

Since we're collecting rounds and automatically player placements, we'd be able to filter easily to top 32, top 16, top 8, etc., to determine what trends are occurring. Do top tiers really help? We can see by filtering the data in this way. With free tournaments and smashfests we can't filter the data.

We can filter the data if the spreadsheets all came from a tournament that would list its location, specific ruleset, entry fee, etc. Filters aren't hard to apply.

The increase of grunt work would be considerable though.

 

[Dr. James Rustles]

All good stuff, I think the one thing I'd like added is stage banned, it's VERY useful to know.

And it'd be cool to keep track of everything else you mentioned to, and probably possible.

I've been around saying I'd do this for months, been working on the system itself for months too, just figured it's about time to lay my claim XD

Do you have anything to show for it yet? Surely you've compiled the the thing at least once.

 

[Overswarm]

Do you have anything to show for it yet? Surely you've compiled the the thing at least once.

I don't think you know how programming works.

 

[metalmonstar]

This is really cool. I was actually thinking about doing something like this a while back for my tournaments. I would leave a notepad by each setup and the players would be required to write down stage, characters, who won, and how long each game took. I figured this would be great as it would help me gauge stage popularity, length of time the event would take, and ensure that how the bracket was updated wasn't based on who came to me first with "results." I think if we present this to TOs as useful for event managing as well as data collecting they will look at it a bit more favorably then if we simply ask them to do more.

Definitely want to make sure time is recorded. One of the more contested items in each Smash game so far is the timer, whether to have it on or off, and if on for how long.

 

[LiteralGrill]

Do you have anything to show for it yet? Surely you've compiled the the thing at least once.

Before I was getting ready to work this for Smash I did something similar by hand for PSASBR to get how the process would work, and we do have the beginnings of a database formed. We need to to do some rearranging probably as we expected to have some features we wont really need, but that shouldn't be too difficult.

All I need is that exact list of things you want recorded, and what you want to be able to search it to find which this thread is giving lots of ideas for that I'm taking down. We may be able to finish faster then you'd imagine with what we have so far already coded.

 

[Dr. James Rustles]

I don't think you know how programming works.

You're right, I don't: In sensible programming, the development cycle is always: 1. Write most of the code first, 2. No real modularity, so you have to unit test in stupid bulk and 3. Go so far with step 2 you give it a first run near completion.

He says he and some other guy have been working on it for a while now and it's all been talk. Surely they have a repo somewhere in the stars that has the program or modules of it partially completed and capable of executing its elementary tasks.

Sometimes it seems like you're only interested in putting other people down to the level you have brought yourself to.

 

[LiteralGrill]

So just a quick observation, where do Melee and 64 fit into this research? It would make sense to take the entire smash community into account.

Just to make sure this is answered, I can easily make the system keep data from every single Smash game. I already was planning on doing so, so expect us to be able to keep data from all these events.

---​

Alright, let me make sure I have what we want here.

Basic Event Information (not listed on slips, just listed for filtered searching)

Event Name
Ruleset
Entry Fee
Payout
Number of Entrants
Date

Country
State
Region

For each match:
Player 1/Player 2
Character Used
Stage Used
Stage banned (if applicable)

Filters to search for:
Well, I could make you able to look spcifically for ALL of this, and things individually by characters I bet as well (E.X. how often a player of a character counterpicks certain stages for example)

I'd LOVE a full list of every possible filter you could imagine for finding results, it'd make the job go MUCH faster. If we can get this done, maybe we can get a few smaller events to test it for us, then if it works launch it up and get it at Apex.

 

[Ulevo]

You might want to throw in necessary information for teams as well.

 

[BSP]

I'm not trolling when I ask this:

What if the most prominent TOs and players just don't care about the data and just roll with what they like? Apparently we had data for brawl, but it didn't do diddly.

 

[Ulevo]

I'm not trolling when I ask this:

What if the most prominent TOs and players just don't care about the data and just roll with what they like? Apparently we had data for brawl, but it didn't do diddly.

We didn't have the amount of data that we're looking to collect here. Even so, assuming that is the case, there is still plenty to learn from the data we would collect. If I wanted to argue, for example, that Marth vs Meta Knight was an even match up, while others argued it was 7-3 in Meta Knights favor, I could use the win ratio % data collected to help in either proving or disproving my points. If we wanted to talk about whether or not a stage is a counterpick versus a neutral, we could see the different win ratios certain characters have as opposed to other stages, and if there is a distinct disparity, make judgments on that. There are plenty of uses for this information.

 

[Zonderion]

I'm not trolling when I ask this:

What if the most prominent TOs and players just don't care about the data and just roll with what they like? Apparently we had data for brawl, but it didn't do diddly.

Regardless of what data you are collecting, there's gotta be people willing to invest the time and effort to make it happen. It's very likely that the prominent TOs won't care. If we can do all we can to make it as easy on them as possible, I believe the success chance would be higher. But even still, of the data we do collect we can still draw analysis from it.

 

[LiteralGrill]

I'm hoping players are just willing to fill out slips, and TOs willing to pay maybe a buck to send em in the mail TOPS if they needed a big envelope. I'd travel to the HUGE events like Apex and such to help with data there if needed for when the events seems like it might be too much, and I bet someone could help rep it up in case I couldn't show.

(If I can finish before Apex and Alex would let us roll there, I'll pack my bags and head for Apex!!!)

Again, all I need to know if what YOU need to know. Then I can create the site so I can enter in data, and create a search option so you can find what you exactly want.

 

[Ulevo]

Well I was thinking that the TO would print the slips for the tourney and hand them to the players for each match they play. The player fills the slip out and hands it to the TO in order to get their results put in to the bracket (that way we're likely to get them filled out). TO collects all the slips and after the tourney sends Capps or whoever else the information via email or PM on the forums. You can accept mail if you wanna but that'd be slower.

 

[LiteralGrill]

Well I was thinking that the TO would print the slips for the tourney and hand them to the players for each match they play. The player fills the slip out and hands it to the TO in order to get their results put in to the bracket (that way we're likely to get them filled out). TO collects all the slips and after the tourney sends Capps or whoever else the information via email or PM on the forums. You can accept mail if you wanna but that'd be slower.

Good by me! I only thought of mail being that scanning in the amount of slips something like Apex might produce would be tough on TOs, smaller events it'd be easy to send it digitally I would expect.

 

[metalmonstar]

Like I mentioned in my earlier post. You will meet much less resistance if you let TOs know why the data is useful for them as well. For example collecting data on stages lets a TO make more localized decisions on what the community wants as a stage list. This helps prevent rule debates during event announcement threads. Also compiling a list of game lengths would give us an average duration. TOs could use this then as a better approximation of tournament duration. Another way to reduce resistance is to make it as easy as possible to collect.

 

[LiteralGrill]

That could easily be done when I release the project, along with its release explain why it is so helpful.

And keeping match times may not be as easy, but if matches are timed out would be. Depends on how aware players are when filling out slips.

 

[Ulevo]

Like I mentioned in my earlier post. You will meet much less resistance if you let TOs know why the data is useful for them as well. For example collecting data on stages lets a TO make more localized decisions on what the community wants as a stage list. This helps prevent rule debates during event announcement threads. Also compiling a list of game lengths would give us an average duration. TOs could use this then as a better approximation of tournament duration. Another way to reduce resistance is to make it as easy as possible to collect.

This thread is merely for the discussion on how to start the project and get it going. Either me or someone else or even a moderator could probably make a sticky thread telling people what the project is and what its for, and why it would benefit our community.

I don't know who you'd talk to about this, but maybe there could be some official TO status or banner you could get if you follow the project and submit slips for x period of time. That would give people some incentive to becoming TO's and follow the project. Do TO's still get blue names/banners?

 

[LiteralGrill]

I don't know who you'd talk to about this, but maybe there could be some official TO status or banner you could get if you follow the project and submit slips for x period of time. That would give people some incentive to becoming TO's and follow the project. Do TO's still get blue names/banners?

That would be awesome, a small logo they could add to events to show they were using the system wouldn't hurt either.

 

[Chiroz]

Capps, I made a rough draft of a Database sketch with almost every possible filter there is.

Apart from the basic ones, with the Database i'll be sending you to your inbox, you can search.

Any game of any Match of any Tournament. Which items were legal on an specific game or every single game where "Bat, Fan and Gun" were legal. Every result between OS's MK and Brood's Olimar with all their information (stage, items, mode, who won, max stocks, stocks left, time left, which stages OS strikes, which ones Brood striked, who picked the stage, etc.). You can also filter by any of the previous information, searching for all of the games where OS has picked the stage RC or all the games where he striked FD.

You can look up every single game between an MK and a DK on an specific stage or with an specific item on. You can look at every single M2K game where he uses Mewtwo and also strikedFD and also won by 2 stocks and there were at least 4 minutes left on the clock.

You could also look for all the Quarter Final games where a Sonic fought a Mario, all the Semi-Finals of tournaments where RC was legal or all the Finals where all 5 games were played and the winner ended with only one stock on the very last game.

You could look up the highest ranking DKs and all of the games they've played with all the information I states previously for every single game. You could also look up specific player/character stats such as M2K's stats using Fox, how many games has he played, how many has he played on FD, how many has he played when the "beam sword" was a legal item, how many has he won, how many times has he fought Ken while using his Fox and how many has he won against Ken while using Fox.

There is a lot more that my Database can filter by than what I have actually written, a lot, a lot more. The Rough Draft I made was analyzed and created in a way where every single element is considered individually as an entity such that you can filter by almost every element that there is in Smash.

If you guys can think of something that wasn't mentioned on this post or mentioned on any other post on this thread (my Rough draft can already do all of the filters asked for in this thread) then list it and I will check if said Rough Draft can do that Query. If it can't I'll fix it.

Capps, I'll be sending you a picture of the Rough Draft (I did it by hand) in 2-3 hours when I can get a hold of the notebook which is at my house.

 

[LiteralGrill]

Database Talk

WOW!

That is some seriously amazing work! I can't wait to see it and start work on getting it done.

COOL NEWS: I'll be talking to Alex Strife about getting this database used at Apex some time soon, we're figuring out a day to talk!

 

[LiteralGrill]

Sorry for the double post, but my programmer has asked how you would like data presented. Any ideas from people?

 

[Ulevo]

Sorry for the double post, but my programmer has asked how you would like data presented. Any ideas from people?

Uh, hm. I figured it would just be presented as requested if it was ever needed, but having monthly reports and stuff would be cool too. I have no idea how it'd be best to show it off though. Anyone else?

 

[LiteralGrill]

Uh, hm. I figured it would just be presented as requested if it was ever needed, but having monthly reports and stuff would be cool too. I have no idea how it'd be best to show it off though. Anyone else?

I was figuring that people could use a search engine to find whatever they wanted data wise themselves, but then it comes down to how you actually show it to people, how should you look at it and how should you show it? graphs, just numbers, what is most useful?

 

[Zonderion]

I would think a combination of both. Probably the best scenario would be to have a website that displays a sort-able, searchable table with the ability to include graphs. I suppose something like Google tables could work if there wasn't a need for its very own website.


Edit: Maybe also with the ability to download the info into a spreadsheet.

 

[Dr. James Rustles]

You can just make the interface natural for inputting data, like a form, and transform the data structures into something that can be packaged into a file format you need. Maybe a tiny interface or API like this exists for OpenOffice. In this way you can make it conducive to Tournament Organizers and data collectors without worrying about having to create your own graphs.

EDIT: If you're doing this just for the filtering options, is there nothing out there that meets the data aggregation our data aggregation needs?

 

[Chiroz]

Capps, I sent you the pictures of the rough draft I made to your inbox.




I believe that the Application User Interface should allow the TO's to introduce this information in the simplest way possible and also allow for the easiest ways to filter searches for those interest in doing data analysis.





Basically, I think that, just like what Quilt said, the "input" side of the application should consist of a multi-level form which would be portrayed as different "steps". Each step would be a different type of information the TO has to input in order to allow the TO to input said data without much thought (simplicity is key).

The first step would consist of the tournament's basic information. Name, Location, Entry Fee, Date, Ruleset (banned stages, allowed items, mode, stocks, time) would be the information asked for in this form. You could have simple textboxes for things such as Name, Entry Fee, etc. Location could be a selection dropbox having the different States or Regions, while Ruleset could be a series of checkboxes (consisting of stages, items and characters) that the TO would check or not depending on whether they are legal or not in his tournament.

The second form would consist of inputting every player's information (Name, Nickname, Location). Every player could do this himself as he is registering for the tournament. Once registration is closed the TO could then proceed to the next level and some information could be inferred from what was introduced (such as number of entrants).

The third level would involve introducing the information of every game in every match. Basically the form would consist of the basic information of a match (Round, Players, etc.) and a series of sub-forms (up to 7) which would consist of the information of each individual game (Characters, stages striked, stage picked, who won, stocks left, time left, comments, etc.). The application would allow the TO to have as many of these forms open as he wants, this way he would be able to open a new form every time a new match starts and slowly fill out each game's sub-form as the games progress. Once a match is finished, the TO would just click on finished and some information could also be inferred. (Such as who won the match).

This last step would continue until the Grand Finals. After this, then all the information is gathered and uploaded to a mirrored database for a validity check. Once it is validated by Capps' staff then it can be uploaded to the real database.





The search/filter side of the application should involve a connection to the real database which cannot overwrite and has locks preventing people who are viewing it to corrupt data mistakenly when the database is being updated.

The User Interface should probably consist of a single form which would allow you to type in many different search filters through the use of textboxes for information such as player names, characters, stages played on, winner of the match, tournament name, tournament date, tournament location, tournament entry fee, tournament entrants. You could also use drop down boxes for filters such as Match Round/Placing (1st Round, Grand Finals, Losers Quarter-Finals , etc.), number of the game on said match (1st game, 2nd game, 3rd game, etc.), stocks left or timer left. And finally have a series of checkboxes to filter which stages were banned, which stages were striked, which items were allowed or which characters were banned. (Each and everyone of these filters on this paragraph should be optional).

After all of this there should be a small sub-form which would ask the user what kind of information he wants to be shown. Such as if he only desires to know how many times Fox has beaten MK, or if he wants to know how many times Fox has beaten MK on each stage individually, or maybe he wants to know all the different players who played Fox and beat MK, or whose Fox has beaten MKs the most on each stage individually, etc. This should be done by giving a series of checkboxes pertaining to each individual information (in this example it would consist of a checkbox titled "Stages", another titled "Players" and his search criteria), each of these checkboxes would open a small sub-menu of checkboxes which would allow you to choose exactly what information you want to be shown (In this example it would consist of a list of all of the stages allowing you to check only the ones you are interested on, or a list of all the Fox player's {this would be displayed dynamically allowing the form to display information from the database in real time} who have beaten an MK and allowing you to check only the ones you are interested on).





After filling out all the previous information on the forms, the application would then perform the search and return the information of each and every game/player/character/item/stage/etc. which matches the search criteria. You could then introduce options such as present information on a graph or anything pretty like that, but its important to keep the core simplicity in order to make it accessible to everyone.

 

[LiteralGrill]

Alright, bringing this thread back as while the boards have been down me and my group have been putting tons of work into a database and site to work with. We're hoping soon we can manage to get something online to work with and find a few TOs willing to test this out, any takers?

 

[Dr. James Rustles]

Alright, bringing this thread back as while the boards have been down me and my group have been putting tons of work into a database and site to work with. We're hoping soon we can manage to get something online to work with and find a few TOs willing to test this out, any takers?

I just asked a local Huntsville, AL organization to look into it. Waiting for a response. Glad to see you guys are still trying this.

 

[LiteralGrill]

I just asked a local Huntsville, AL organization to look into it. Waiting for a response. Glad to see you guys are still trying this.

We aren't giving up, we really are doing quite well on it. We'll just have to get everyone using it.

 

[Zonderion]

It has been unfortunate that Smashboards has been down, and I am so thankful they are back up. But as Capps said, we have have made a lot of progress during this down time.