How to write a for loop that creates a vector of [1:1000, 1001:2000,....,n]

I am trying to write a loop that creates a vector with blocks of data. How exactly would the above be possible?

[Jon]

I can not understand your question. The expression you have would just expand to a vector which starts at 1 and increments by one all the way up to n. There is no jump between the lastt element of the first group, which is1000 and the first element of the next group which is 1001.

Maybe you could further explain what it is you are trying to do, perhaps with a simple example of what you'd like your code to do

For future reference, it is better style for the MATLAB answers to have the subject of your answer just describe the problem in words, and then provide details in the body.

[Kevin Egan]

Thanks,

Essentially I would like to create a vector of blocked data that goes up to 100000. So the first block would be 1:1000, the next block would be 1001:2000, and continue with this process until 100000. In general, I was hoping for a way to create a vector like this without having to manually code the vector.

[Kevin Egan]

Eventually, I would rearrange this vector and apply it to a dataframe to allow me to rearrange that dataframe similar to methods used in block bootstrap.

[Geoff Hayes]

Kevin - do you need to use a loop or could you just use reshape?

[the cyclist]

We don't know what you mean by "blocked data". Suppose you only wanted "blocks" of 3. I guess you somehow mean

v = [1:3 4:6 7:9 10:12];

But how -- specifically -- is that different from

v = 1:12;

which you can see in MATLAB give exactly the same result for the contents of v.

Can you write out in MATLAB code what the finally result should be, in this small example? Maybe you actually want "blocks" as different rows, like

v = [1:3; 4:6; 7:9, 10:12];

which is a 4x3 matrix?

[Kevin Egan]

reshape works, do you know if it is possible to randomize the order of the results and create a vector in the process?

[Geoff Hayes]

by randomize the order of the results do you mean that rather than blocks of 1:1000, 1001:2000, etc. you want the 1:100000 randomly "distributed" across the 100 blocks?

[Kevin Egan]

Sorry again, I was hoping for something like [1:1000, 1001:2000, 4001:5000, 2001:3000....]

However, I think I was able to do that manually. Thanks for the help with reshape.

Answer 1

[the cyclist]

% Define block size and number of blocks
blockSize = 1000;
numberBlocks = 50;

% Create vector of all values
A = 1:(blockSize*numberBlocks);

% Reshape into blocks
A = reshape(1:(blockSize*numberBlocks),blockSize,numberBlocks);

% Randomize the blocks
A = A(:,randperm(numberBlocks));

% Convert back to vector
A = A(:)';

Some of these steps could be combined, but I separated them for clarity.

I recommend testing this with a very small block size (maybe 3) and number of blocks (maybe 5) to verify it does what you want.

[Guillaume]

Another way to obtain the same:

blockSize = 1000;
numberBlocks = 50;

A = reshape((randperm(numberBlocks) - 1) * blockSize + (1:blockSize)', 1, [])

Answer 2

[Jon]

Just to make Geoff's suggestion more concrete, suppose you had a length 100000 vector of data points. You could put this into a matrix with 2000 rows and 50 columns. Each column would be a "block" of data if this is what you mean. To do this just use

X = reshape(x,2000,50)

Note, I still don't really understand what you are trying to do, but just to match up a little closer to your description if the vector x in the above expression went from 1:100000 then the first column of X would have the numbers 1 up to 1000 the next column would have the number 1001 up to 2000 etc