Chief Delphi - 2009 Division Algorithm

Chief Delphi (http://www.chiefdelphi.com/forums/index.php)

- Championship Event (http://www.chiefdelphi.com/forums/forumdisplay.php?f=12)

- - 2009 Division Algorithm (http://www.chiefdelphi.com/forums/showthread.php?t=76632)

Schnabel

08-04-2009 22:08

2009 Division Algorithm

So has anyone figured out this year's algorithm? I haven't taken the time to figure it out yet.

jennifer_1547

08-04-2009 22:09

Re: 2009 Division Algorithm

neither have i
it looks so random to me though :|

waialua359

08-04-2009 22:12

Re: 2009 Division Algorithm

It was handpicked by Sean and Dave. :P :P :P LOL!

David Brinza

08-04-2009 22:16

Re: 2009 Division Algorithm

Write the team numbers in order on 348 index cards, shuffle three or four times, then deal them into four piles.

Or you could write a program to do the same thing...;)

MrForbes

08-04-2009 23:06

Re: 2009 Division Algorithm

they used the same random number generator we used for our autonomous drive code

(oops did I say that?)

The Lucas

08-04-2009 23:42

Re: 2009 Division Algorithm

Quote:

Originally Posted by David Brinza (Post 847866)

Write the team numbers in order on 348 index cards, shuffle three or four times, then deal them into four piles.

Or you could write a program to do the same thing...;)

Similar to what I was thinking but slightly different. Your method probably wouldn't produce as much clumping as mine.

Sunbun

08-04-2009 23:42

Re: 2009 Division Algorithm

I'm guessing they used something like this?

Or this?

Chris is me

09-04-2009 01:39

Re: 2009 Division Algorithm

Being completely serious I think they just took 4 teams at a time, randomly sorted them, then moved onto the next 4 teams, etc.

GaryVoshol

09-04-2009 07:14

Re: 2009 Division Algorithm

It was not sorted by team number, team location or team name. I tried plotting all three to see if there was a pattern, and I didn't find one. Someone else posted that it wasn't based on registration date, based on how the last registrants are distributed.

So unless they found some other sorting mechanism, maybe based on some performance criteria such as quals ranking, it appears to be a random assignment. Random would give you strings of teams all assigned to the same division, such as 236 through 271 all being in Curie.

JackG

09-04-2009 10:50

Re: 2009 Division Algorithm

On a hunch, I added together the team numbers from each division to see if there was some pattern. Perhaps the total team numbers might be the same for all divisions?

Team No. Total
Archimedes: 116639
Curie: 110256
Newton: 106945
Galileo: 103721

Apparently, there seems to be no pattern. :(

jennifer_1547

09-04-2009 10:51

Re: 2009 Division Algorithm

i still think that it was done completely randomly
im sticking by it until someone proves what they did to figure it out :)

Andrew Schreiber

09-04-2009 10:52

Re: 2009 Division Algorithm

Has anyone looked at age of team? Win/Loss Record? OPR/DPR?

Based purely on team number the graph appears somewhat sinusoidal (albeit with pretty bad resolution)

Of course, there is always the chance that it is random which would ruin all fun.

JesseK

09-04-2009 12:52

Re: 2009 Division Algorithm

Code:

private List<Team> mTeamList = new ArrayList<Team>();

private final static int sNUM_DIVISIONS = 4;

private static Timer t = new Timer("Randomizer");

private static TimerTask divTask = new TimerTask(){

    Integer division = -1;



    @Override

    public void run(){

       synchronized(division){

           division = (int)(new Random(System.nanoTime()).nextDouble() * sNUM_DIVISIONS);  

           //Indexes the divisions from 0 to 3, which would match the ordinals of division enums

       }

    }



    public int getDivision(){

       synchronized(division){

           return division;

       }

    }

}



static{

   initTeamList();  // Populates the team list with teams who are attending Atlanta

   t.schedule(divTask, 0, Math.random()*1000);

   Iterator<Team> it = mTeamList.iterator();

   Timer teamDivTimer = new Timer("Division Assignment");

   TimerTask teamTT = new TimerTask(){

       public void run(){

          if(it.hasNext())

             it.next().setDivision(divTask.getDivision();

          else

             cancel();

       }

   }

   teamDivTimer.schedule(teamTT, 0, Math.random()*1000);

}

Run it, the select random teams to even out the number of teams in each division. 'nuff said

dodar

09-04-2009 12:54

Re: 2009 Division Algorithm

Ok how about that, but in Lamen's terms

Nick Lawrence

09-04-2009 12:56

Re: 2009 Division Algorithm

I like how 1114, 2056 and 1503 are in the same division.

AdamHeard

09-04-2009 15:29

Re: 2009 Division Algorithm

Quote:

Originally Posted by Nick Lawrence (Post 848080)

I like how 1114, 2056 and 1503 are in the same division.

I like it only because it's not my division :D

R.C.	09-04-2009 15:58

Re: 2009 Division Algorithm

Quote:

Originally Posted by AdamHeard (Post 848184)

I like it only because it's not my division :D

Agreed :yikes:

Gaurav27

09-04-2009 20:48

Re: 2009 Division Algorithm

Quote:

Originally Posted by JackG (Post 848031)

I did something similar; an average of the team numbers in each division.

Team No. Avg.
Archimedes: 1279.42
Curie: 1074.90
Newton: 1062.10
Galileo: 1231.96

I agree there's no pattern here. :confused:

Quote:

Originally Posted by Nick Lawrence (Post 848080)

I like how 1114, 2056 and 1503 are in the same division.

Go Canada! Sorry for the bias. :p

Nick Lawrence

09-04-2009 21:21

Re: 2009 Division Algorithm

Bias for the win.

TKM.368

10-04-2009 17:00

Re: 2009 Division Algorithm

Decided to see if it had anything to do with where teams originated from.

Code:

Ori Arc Cur Gal New Tot

AL    0   1   0   0   1

AR    0   0   0   1   1

AZ    2   3   1   2   8

BRA   0   0   1   0   1

CA    9   8   7  10  34

CAN   5   4   0   3  12

CO    0   0   1   2   3

CT    3   3   4   5  15

DE    0   0   0   1   1

FL    7   1   4   3  15

GA    4   1   1   2   8

HI    0   1   2   1   4

IA    0   0   1   0   1

ID    0   1   0   1   2

IL    1   1   3   1   6

IN    2   2   4   3  11

ISR   0   1   2   2   5

KS    0   1   1   0   2

KY    0   0   0   1   1

LA    0   1   0   0   1

MA    3   3   0   3   9

MD    0   1   1   4   6

MEX   0   0   0   1   1

MI    9  16   9   9  43

MN    3   1   0   1   5

MO    4   2   0   3   9

MS    1   2   0   1   4

NC    1   0   2   0   3

NH    1   2   3   1   7

NJ    5   4   9   2  20

NV    1   1   1   0   3

NY    3   8   8   9  28

OH    3   1   2   1   7

OK    3   0   2   2   7

OR    1   2   3   0   6

PA    6   1   2   0   9

PHI   0   1   0   0   1

PR    0   1   1   0   2

RI    0   0   0   2   2

SC    1   2   1   0   4

TN    0   0   1   0   1

TX    5   3   5   1  14

VA    0   2   3   6  11

WA    2   1   2   1   6

WI    2   3   0   3   8

WV    0   1   0   0   1

Tot  87  87  87  88 349

Guess not...

Alex Golec

10-04-2009 18:16

Re: 2009 Division Algorithm

If I may suggest one other factor, for your entertainment:

Perhaps it is by time of registration, in a 1-2-3-4 fashion?

I say this because one fateful day, back in whenever (2005-2007ish), the team list for the championship was unsorted and showed the order of signing up... teams like 47, 16, 190 were up top.

Anyways, that data is (as far as I know) not viewable.

-Alex "twocent" Golec

XXShadowXX

10-04-2009 19:11

Re: 2009 Division Algorithm

I'm gonna say random. So i can get sleep tonight.

Nawaid Ladak

10-04-2009 23:58

Re: 2009 Division Algorithm

Added totals of each teams Lat/Long Location? in a serpentine order?

ScottOliveira

11-04-2009 00:01

Re: 2009 Division Algorithm

Quote:

Originally Posted by Nawaid Ladak (Post 848852)

Added totals of each teams Lat/Long Location? in a serpentine order?

Have fun calculating that....

JackG

11-04-2009 22:12

Re: 2009 Division Algorithm

Quote:

Originally Posted by Alex469 (Post 848740)

Perhaps the teams who qualified from the Michigan State Championship would serve as a test group. If that really is the formula used, then theoretically, they'd be divided evenly among the divisions.

Mike Schreiber

13-04-2009 20:57

Re: 2009 Division Algorithm

Quote:

Originally Posted by JackG (Post 849326)

Perhaps the teams who qualified from the Michigan State Championship would serve as a test group. If that really is the formula used, then theoretically, they'd be divided evenly among the divisions.

Not necessarily, I know for a fact we were already registered for Atlanta before we qualified, same with 217 and 33 and I'm sure there were many others.

Alex Golec

13-04-2009 21:14

Re: 2009 Division Algorithm

Quote:

Originally Posted by JackG (Post 849326)

Perhaps the teams who qualified from the Michigan State Championship would serve as a test group. If that really is the formula used, then theoretically, they'd be divided evenly among the divisions.

While there were 18 teams that qualified that day, I don't know how many of them accepted the invitation. In addition, there were another (up to) 12 qualifiers at the two Minnesota events that could skew this, along with the opening of the wait list if there were more spots available at that time.

I suppose we'll never truly find out.

Jared Russell

14-04-2009 07:37

Re: 2009 Division Algorithm

Quote:

Originally Posted by Alex469 (Post 848740)

Checked it based on the last handful of teams (whose registration order we know from watching the CMP "who's going" page on the FIRST site). No (simple) pattern to be found.

The Lucas

15-04-2009 01:23

Re: 2009 Division Algorithm

1 Attachment(s)

I know I said it was "clumpy" (6 teams in a row) but I did some analysis and the 6 in a row should happen ~1/3 of the time in random (with 348 teams). Other than the 6 in a row it fits well to expected (spreadsheet attached). I am going to say it was simply random.

All times are GMT -5. The time now is 04:39.