How Do I use nested for loops to first go through every column, and then do a calculation for every row of that column.

I have a data set that is about 10,000 rows and 51 columns. I’m trying to create nested for loops where it will go through every row of the current column, do a calculation, and then move to the next column and replace all the original values with the new ones. This might be simple but I’m still a beginner to coding but I just can’t get this to work right.

[David Fletcher]

It's unlikely you would need a nested loop. More likely you would iterate over the columns, recalculate each column, and then replace the entire next column with the recalculated values. Without knowing more about your specific problem, it's hard to say for certain though.

[Clay Turpin]

I am not sure that this would work that way. This is just part of a giant algorithm and for it, each column is essentially its own data set. So, I basically need to go down the rows each column, and do a calculation and just replace the old data in that row with the new calculated values.

[David Fletcher]

Yeh, that may well be the case. Though without knowing about your exact problem in greater detail, you're unlikely to get any form of definitive answer (if that is indeed what you are seeking)

[Clay Turpin]

matrix = csvread('M3_Data_HeatingTimeHistories2.csv');

[rows,ncolumns] = size(matrix); % finds size of matrix

xdata = matrix(:,1);

ydata = matrix(:,2:end);

counter = 0;

for column = (1:1:ncolumns)

     counter = counter + 1;

          for row = (9:1:rows-9) 

               ydata(row,column) = (matrix(row+8,column) + matrix(row-8,column) + matrix(row+7,column) + matrix(row-7,column) + matrix(row+6,column)+ matrix(row-6,column)+ matrix(row,column)+ matrix(row+2,column)+ matrix(row-2,column)+ matrix(row+4,column) +matrix(row+5,column) + matrix(row-5,column) + matrix(row-4,column) + matrix(row+3,column)+matrix(row-3,column) + matrix(row-1,column) + matrix(row+1,column)) / 17;

         end

end

smooth_data = [xdata ydata];

Here is my code, what it should be doing is taking 8 values before, and 8 values after, and averaging them and replacing the current number with the new calculated value. Its basically smoothing the curve that is given in the file I am loading.

[David Fletcher]

You may find you get more interest if you format your code so it's more readable (a bit like you've done with that little block). But from what you describe it sounds like a moving mean on each column see the movmean function

[Clay Turpin]

I'm sorry this is my first time using mathworks. We are not allowed to use any functions, we have to manually calculate and smooth the data ourselves.

[David Fletcher]

That's one hell of a calculation you've got in the for loop - you must have got RSI just typing it out. Have a look at the answer I've put below and see if it works OK (and you understand everything that is happening)

[Brendan Hamm]

A couple of quick points:

1) Without using functions, you do have the right approach at a nested loop.

2) for the rows loop, you can actually go all the way to rows-8, that is (9:1:rows-8)

3) You may not want to start at column = 1, but rather column = (2:1:ncolumns) as it appears that column 1 is the xdata.

4) Do you also want to average the data in rows 1 through 8, etc. but using less data points? if so, consider padding with zeros to perform the calculations.

5) Is there any point of confusion you had? It seems you are doing what you set out to. Presuming the code you wrote is doing what you want.

Answer 1

[David Fletcher]

matrix = csvread('M3_Data_HeatingTimeHistories2.csv')  
[rows,ncolumns] = size(matrix); % finds size of matrix

    xdata = matrix(:,1);
    ydata = matrix(:,2:end);

    window=8

    for column = (1:(ncolumns-1))
              for row = (window+1:rows-window) 
                   ydata(row,column)=sum(ydata(row-window:row+window,column))/(2*window+1)
              end
    end

[Clay Turpin]

Sorry this was so poorly explained lol. But thank you for all your help! I understand it all now!