Selecting very specific data from a table using a for loop

Hi struggling with some simple Matlab, I have a 49440x3 table created from excel data, with columns 'title' 'x' 'y', the 'x' column contains values starting at 0.25 incrementing by 0.25 until it reaches 154.5 in a loop. I want to put each set of 0.25:0.25:154.5 into separate tables with the associated 'title' and 'y'.

Using a for loop I tried:

    vars={'title','x','y'};
    n=1;
    for n=n+618;
        p=n+617;
        T=leafdata(n:p,vars)

    end

This obviously overwrites the new table because I dont know how to create a dynamic variable instead of 'T' I tried:

   vars={'title','x','y'};
   n=1;
   x=1;
   for n=n+618;
       p=n+617;
       T(x)=leafdata(n:p,vars)
       x=x+1;
   end

And also

    vars={'title','x','y'};
    n=1;
    x=1;
    for n=n+618;
        p=n+617;
        x=leafdata(n:p,vars)
        x=x+1;
end

Neither work and I have also tried starting the for loop with

vars={'title','x','y'};
n=1;
for rows=leafdata.x
  if rows==0.25
  n=n+1;
  end
  n=leafdata(rows,vars)
end

I would prefer to use a loop that searches for the 0.25 if possible? Or make T change to T1, T2...Tn after each loop. But Any solutions to this would be very welcome!

Answer 1

[dpb]

"I want to put each set of 0.25:0.25:154.5 into separate tables with the associated 'title' and 'y'."

I'd ask "why"? It's a very typical newbie idea in Matlab to build more variables (in this case tables) but when they're associated it's rarely really the best solution. I'd suggest using a 2D array instead and address the sequences by column.

N=154.5*4;  % number in each set
leafdata.x=reshape(leafdata.x,N,[]);

will do the rearrangement; use same for the other variables. Then each group is addressed as (:,nGrp) for subscripting expression for arbitrary group nGrp.

"Using a for loop I tried:"

n=1;
for n=n+618;
  p=n+617;
  x=leafdata(n:p,vars)

The above does nothing; n+618 is simply a single value of 619 and p=618 so you try to address [619:618] which is an empty set since the second array index is less than the first. But, it's a nonproductive path down which to head as addressed by the last question...

"Or make T change to T1, T2...Tn after each loop"

This is the subject of an FAQ that basically says "don't do that!" --

http://matlab.wikia.com/wiki/FAQ#How_can_I_create_variables_A1.2C_A2.2C....2CA10_in_a_loop.3F

[Matt Jolley]

The reason I want to split the data is because this has been collected from an electromechanical machine with 80 motors and each has been sampled from 0.25s to 154.5s and each y value stored in the next column is the position. The data is stored in the same three columns for all 80 motors. I want to plot each motors movement seperately and because I dont know how to plot the data seperately if its all in one table I wanted to seperate it and plot the data from individual tables. Any further hints?

Answer 2

[dpb]

Sure...as said, if you do the reshape to columns, you'll end up with a column of T, X, and Y for each machine/motor. That's the most logical arrangement you could have -- to plot any given motor you simply write

mtrNum=input('Which motor do you wish to plot?');
plot(T(:,mtrNum), [X(:,mtrNum) Y(:,mtrNum)])
legend('X','Y')
xlabel('Time')
ylabel('Position')
title(num2str(mtrNum,'Motor %d Position vs. Time'))

ADDENDUM

NB: If the data are indeed collected at the same time for each group then you can simplify the final dataset further by keeping only one time vector as they're all redundant.

T=T(:,1);      % keep only the one time vector

Then you can eliminate the subscripting for it everywhere it's referenced--

plot(T, [X(:,mtrNum) Y(:,mtrNum)])

Also, to show further the power of keeping such an arrangement instead of many variables, to get (say) the distance the motor has traveled--

[~,d]=bsxfun(@minus,cart2pol(X,Y),cart2pol(X(1,:),Y(1,:)));

for all 80 (or however many there are) at "one swell foop". bsxfun is a builtin function that does singleton expansion for vector/array operations. See

doc bsxfun % for the gory details