Several Operations Within a Matrix Matlab

i'd like someone with knowledge on matlab to help me sort this out, the plan is to write a script as follow

it starts reading numbers on column 2 and check in column 3 one if they are repeated: The Condition is, if a number in column 2 is repeated more than one time in column 3, the corresponding values in column 4 will be summed up, if a number from column 2 is not repeated or only repeated one time in column 3, its corresponding value in column 4 will be left as it is. The obtained values would be generated in a vector. the letters were put there just for explanation here is an example to try with

data=[0   1    0     0
      1   2    1    200
      2   3    1    300
      3   4    1    400
      4   5    2    500
      5   6    3    600
      6   7    4    700
      7   8    5    800
      8   9    5    900
      9   10   5    1000
      10  11   6    1100
      11  12   7    1200
      12  13   7    1300];

this would be the desired result

v=[(200+300+400) 200 300 400 (800+900+1000)  600 (1200+1300)  800 900 1000 1100 1200 1300]

waiting on any ideas.Thank you programmers !

Answer 1

[Stephen23]

data=[0   1    0     0
      1   2    1    200
      2   3    1    300
      3   4    1    400
      4   5    2    500
      5   6    3    600
      6   7    4    700
      7   8    5    800
      8   9    5    900
      9   10   5    1000
      10  11   6    1100
      11  12   7    1200
      12  13   7    1300];
D = diff(data(2:end,3),1);
E = [false;~D] | [~D;false];
tmp = cumsum(diff([false;E])>0);
out = data(:,4);
idx = data([false;diff([false;E])>0],3);
out(idx) = accumarray(tmp(E),data([false;E],4));

And this code's output compared to your example output:

>> [out,v]
ans =
         900         900
         200         200
         300         300
         400         400
        2700        2700
         600         600
        2500        2500
         800         800
         900         900
        1000        1000
        1100        1100
        1200        1200
        1300        1300

[sami elahj]

Stephen your code worked for thank you :)

Answer 2

[dpb]

Doesn't look to me like the result is really dependent upon the third column at all; it's simply accumulating the like values in the fourth....

 >> v=accumarray(data(:,3)+1,data(:,4))
v =
           0
         900
         500
         600
         700
        2700
        1100
        2500
>>

*NB: * the need to add one to the column values in column three to be valid array indices of positive integers.

[sami elahj]

>> v =

           900
           200
           300
           400
          2700
           600
          2500
           800
           900
          1000
          1100
          1200
          1300

hey dpb this would be the desired result, i don't understand your note at the end

Answer 3

[Guillaume]

Your output vector is really weird. Are you sure that's really what you want?

Ir may be possible to achieve it without a loop but the resulting code may be very obscure. Anyway, here is one way of doing it:

[values, ~, subs] = unique(data(:, 3));  %it's much safer to use unique than adding 1 to dat(:, 3)
valsum = accumarray(subs, data(:, 4)); %since it works regardless of the range of values
count = accumarray(subs, 1);
v = zeros(size(data(:, 3)));
for row = 1:numel(v)
   valrow = values == data(row, 2); %empty if data(row, 2) is not found in column 3
   if count(valrow) > 1  %note that empty>1 returns false, so this works properly 
       v(row) = valsum(valrow);
       count(valrow) = 1;  %prevents current valsum to be included again.
   else
       v(row) = data(row, 4);
   end
end

[sami elahj]

@Guillaume i tried your code but i got this thanx anyway

[Guillaume]

An obvious typo that you should have spotted. Replace counts by count.