CODEWITHSUNDEEP
matlab
yankeefan11
yankeefan11

Reputation: 485

Calculate Ranking from previous occurrence in matlab

I am working with trying to do some rankings for my chess club. I figured this would be a good chance to work on coding to find ELO. The way I encode it is each year we reset the ranking to 1200 then calculate it. So I have the vectors year, winner, loser and I want to make elo. Winner and losers are just numbers 100-125 to identify person and I have 3 years of data.

So I am struggling with doing this correctly. My struggle is figuring out how to find the most recent score since when I look through each row, I need to search for the last time someone’s ID was called in either win or lose column. Additionally, if year changes, I need to reset the ELO score (I think using unique(year) should do).

Here is what I have tried (the disp('here') is just to show me if it runs, which it never does): 

for i = 1:length(Season2014toNow);
WID = WinteamID(i);
LID = LoseteamID(i);
Seasoni = Season2014toNow(i);

for j = i-1:-1:1;
    if WID == WinteamID(j) & Seasoni == Season2014toNow(j);
        Win_old_team_elo(i) = Win_new_team_elo(j);
        disp('here')
        break
    elseif WID == LoseteamID(j) & Seasoni == Season2014toNow(j);
        Win_old_team_elo(i) = Loss_new_team_elo(j);
        disp('here')
       break
    else 
        Win_old_team_elo(i) = 1200;
        break
    end
end

for j = i-1:-1:1;
    if LID == WinteamID(j)& Seasoni == Season2014toNow(j);
        Loss_old_team_elo(i) = Win_new_team_elo(j);
        break
    elseif LID == LoseteamID(j& Seasoni == Season2014toNow(j));
        Loss_old_team_elo(i) = Loss_new_team_elo(j);
        break
    else
        Loss_old_team_elo(i) = 1200;
        break
    end
end    


[Win_new_team_elo(i),Loss_new_team_elo(i)] = CalcElo(Win_old_team_elo(i),Loss_old_team_elo(i)) ;
end

Upvotes: 0

Views: 42

Answers (1)

Mark_Anderson
Mark_Anderson

Reputation: 1324

The code breaks for i=1, which is the first step in the outermost for loop

 for i = 1:length(Season2014toNow);

for i=1, we get a i-1=0 in the inner loops

for j = i-1:-1:1;

which is the impossible instruction "start at zero and subtract one until we get to 1".

One route around this is to add an exception for i=1 that allows us to skip this code that breaks the programme. eg.

for i = 1:length(Season2014toNow);
WID = WinteamID(i);
LID = LoseteamID(i);
Seasoni = Season2014toNow(i);

%% New if statement
if i==1 %%initialise scores and skip the code breaking i-1 loop
   Win_new_team_elo(i) = 1200;
   Loss_new_team_elo(i)= 1200;

else %% do the preexisting code 

  for j = i-1:-1:1;
     if WID == WinteamID(j) & Seasoni == Season2014toNow(j);
        Win_old_team_elo(i) = Win_new_team_elo(j);
        disp('here')
        break
    elseif WID == LoseteamID(j) & Seasoni == Season2014toNow(j);
        Win_old_team_elo(i) = Loss_new_team_elo(j);
        disp('here')
       break
    else 
        Win_old_team_elo(i) = 1200;
        break
    end
  end

  for j = i-1:-1:1;
    if LID == WinteamID(j)& Seasoni == Season2014toNow(j);
        Loss_old_team_elo(i) = Win_new_team_elo(j);
        break
    elseif LID == LoseteamID(j& Seasoni == Season2014toNow(j));
        Loss_old_team_elo(i) = Loss_new_team_elo(j);
        break
    else
        Loss_old_team_elo(i) = 1200;
        break
    end
  end    


  [Win_new_team_elo(i),Loss_new_team_elo(i)] = CalcElo(Win_old_team_elo(i),Loss_old_team_elo(i)) ;

end %% closing the if statement I added
end

A second route around this would be to use the find function. If my understanding of your code is correct you are aiming to find the first occurrence of WID==WinTeamID in the current year

%% Mask for current year
AllWinTeamID_ThisSeason = WinteamID(Season2014toNow==Seasoni)

%% Find the first ID that matches
% SYNTAX (condition, #items to return, order)
find(AllWinTeamID_ThisSeason== WID, 1,'first')

Which can be built into something to replace the j-loops in your code

PS. I can't see your CalcElo code. I'm assuming it works and is correct.

PPS. Other minor fixes to get to best practice MATLAB programming would be to indent code for legibility (highlight in MATLAB and press Ctrl+i / Cmd+i), add comments with %% to explain what bits of code are doing, and to put your j-loops into a function and call that function, rather than have the same code copied out twice.

Upvotes: 1

Related Questions