Splitting an array every n rows

Hi everyone,

fairly new at matlab and i am having a problem with arrays.

My goal: I have an array(mx3) of data that i want to separate it every N rows so i can save that (sub)array as a separate file.

I am having trouble with how the loop should be setup:

p=1;
n = ceil(length(Accell)/N); % number of samples
y=N;
for i = 1:n 
    for q = 1:j 
        M{i,q} = Accell(p:y,q);  
    end
end

A obvious problem occurs; the variable p and y stay constant so the row won't change. Accell is my Data en N is the number of rows I want to separate.

so ideally M would be:

M{1,:} = Accel(1:80,:)

M{2,:} = Accel(81:160,:)

etc etc

After this loop I extract the rows of M into a array and save it.

    for x=1:size_M % Number of samples (rows) in M
        i=1;
        while exist(['Saved Data\', num2str(classificatie), '\', num2str(i.', 'D%03d'), '.mat'], 'file')==2
        i=i+1;
        end
        inputs = [M{x,1},M{x,2},M{x,3}];  
        save((['Saved Data\', num2str(classificatie), '\', num2str(i.', 'D%03d'), '.mat']), 'inputs')
        % set(handles.debug, 'string','Measurement values have been saved successful')
    end

Right now the code saves the first 80 rows of the data (Accell) in n (samples) files. I know it's very convoluted and I am very open to improvements.

Thank you in advance!

-Vincent

[KALYAN ACHARJYA]

Can you elaborate your problem with some example?

Input (consider example as representation) and what result you expect?

[Vincent den Ronden]

Ofcourse. I am making a model using a mobile phone as accelerometer.

I will measure for a set time t.

This gives me a array of data with m rows and 3 colums.

I want to split this array every N rows and save that new array.

input is the attachement .mat file.

The result is mutliple files (see picture) each containing rows of the input array.

The story is that every 4 seconds we want to do some feature extraction of the signal so we can use it as input to a model.

Thank you for you fast resonse!

edit: The problems I am having are that the index can exceeds the dimension. Since the input data isn't guarenteed to be a multiple of rows N.

Answer 1

[Stephen23]

"so ideally M would be:"

M{1,:} = Accel(1:80,:)
M{2,:} = Accel(81:160,:)
etc etc

Just use mat2cell:

>> N = 80;
>> A = rand(192,3); % fake data, # of rows != multiple of N.
>> S = size(A);
>> R = N*ones(1,ceil(S(1)/N)); % number of rows in each cell.
>> R(end) = 1+mod(S(1)-1,N); % number of rows in last cell.
>> M = mat2cell(A,R,S(2));

And checking:

>> cellfun('size',M,1)
ans =
   80
   80
   32
   

Or, if you really want to use a loop, you can use end inside the array indexing:

R = ceil(S(1)/N);
M = cell(R,1);
for k = 1:R
    B = (k-1)*N+1;
    E = k*N;
    M{k} = A(B:min(end,E),:);
end

And checking:

>> cellfun('size',M,1)
ans =
   80
   80
   32

[Vincent den Ronden]

Thank you so much! Exactly what i meant! Didn't even know the function existed.