Problem with matrix manipulation- how can I avoid having to type in everything?

I am looking to extract data from matrices, training (1500 by 3) and testing (500 by 3), n is the number before the colon. this is what I have by far;

% for n = 1
newTrainInput = training(1:1500,2);
newTrainOutput = training(1:1500,3);
newTestInput = testing(1:500,2);
newTestOutput = testing(1:500,3);

%for n=2
newTrainInput = [training(2:1500,2),training(1:1499,2)];
newTrainOutput = training(2:1500,3);
newTestInput = [testing(2:500,2),testing(1:499,2)];
newTestOutput = testing(2:500,3);

%for n=3
newTrainInput = [training(3:1500,2),training(2:1499,2),training(1:1498,2)];
newTrainOutput = training(3:1500,3);
newTestInput = [testing(3:500,2),testing(2:499,2),testing(1:498,2)];
newTestOutput = testing(3:500,3);

%for n=4
newTrainInput = [training(4:1500,2),training(3:1499,2),training(2:1498,2),training(1:1497,2)];
newTrainOutput = training(4:1500,3);
newTestInput = [testing(4:500,2),testing(3:499,2),testing(2:498,2),testing(1:497,2)];
newTestOutput = testing(4:500,3);

%for n=5
newTrainInput = [training(5:1500,2),training(4:1499,2),training(3:1498,2),training(2:1497,2),training(1:1496,2)];
newTrainOutput = training(5:1500,3);
newTestInput = [testing(5:500,2),testing(4:499,2),testing(3:498,2),testing(2:497,2),testing(1:496,2)];
newTestOutput = testing(5:500,3);
-
-
-
%for n=10
newTrainInput = [training(10:1500,2),training(9:1499,2),training(8:1498,2),training(7:1497,2),training(6:1496,2),training(5:1495,2),training(4:1494,2),training(3:1493,2),training(2:1492,2),training(1:1491,2)]; 
newTrainOutput = training(10:1500,3);
newTestInput = [testing(10:500,2),testing(9:499,2),testing(8:498,2),testing(7:497,2),testing(6:496,2),testing(5:495,2),testing(4:494,2),testing(3:493,2),testing(2:492,2),testing(1:491,2)];
newTestOutput = testing(10:500,3);

Answer 1

[Geoff Hayes]

Jessie - I think tha thou can simplify the above to something like the following

 for u=1:n

    newTrainInput = [];
    newTestInput  = [];

    for v=u:-1:1
        newTrainInput  = [newTrainInput , training(v:1500-(u-v),2)];
        newTestInput   = [newTestInput  , testing(v:500-(u-v),2)];  
    end

    newTrainOutput = training(u:1500,3);
    newTestOutput  = testing(u:500,3); 

    % do something with the arrays

 end

In your example, n would be 10 and we would then have an inner loop where we would append data to the newTrainInput and newTestInput as per your above description. I haven't tested this, but think that it will provide a start in simplifying your code.

[Jessie Banks]

Thank you it worked. But it produces a warning about the variable changing size on every loop iteration and suggests preallocating for speed.

[Geoff Hayes]

The warning is expected. If you want to avoid it, try allocating memory to the newTrainInput and newTestInput arrays before you enter the inner for loop. Perhaps something like

 newTrainInput = zeros(1500-(u-1),2*(u-1));

Then the code to update this array would have to change so that you access blocks within it on each iteration of the for loop. Something like

 atBlock = 1;
 for v=u:-1:1
     newTrainInput(:,atBlock:atBlock+1) = training(v:1500-(u-v),2);
     atBlock = atBlock + 2;
 end

[Geoff Hayes]

Jessie's answer moved here

I got an error in line 9 : newTrainInput(:,atBlock:atBlock+1) = training(v:1500-(u-v),2); I used the code;

for u=1:n
      newTrainInput = zeros(1500-(u-1),2*(u-1));
      newTestInput = zeros(500-(u-1),2*(u-1));
      atBlock = 1;
   for v=u:-1:1
       newTrainInput(:,atBlock:atBlock+1) = training(v:1500-(u-v),2);
     atBlock = atBlock + 2;
      newTestInput(:,atBlock:atBlock+1) = testing(v:500-(u-v),2);
     atBlock = atBlock + 2;
   end
      newTrainOutput = training(u:1500,3);
      newTestOutput  = testing(u:500,3); 
      % do something with the arrays
end

[Geoff Hayes]

Jessie - please include the error message. As well, try stepping through the code with the debugger to see what is happening.