How to group values in a vector based on certain condition?

I have a row vector that has values 1 to 5 that keeps repeating like a cycle as following.

row = [1; 2; 3; 4; 5; 1; 1; 2; 3; 4; 5; 5; 1; 1; 2; 2; 3; 4; 5; 5]

I need to group this data taking 1 and 5 as reference. If the 1 or 5 is occuring more than once consecutively, consider the first 1 and last 5.

Ans =

[1 2 3 4 5 0 0 0;

1 1 2 3 4 5 5 0;

1 1 2 2 3 4 5 5 ]

Any ideas?

Thanks

Answer 1

[Voss]

If the sub-sequence always starts with a 1 and ends with a 5, you could do this:

row = [1; 2; 3; 4; 5; 1; 1; 2; 3; 4; 5; 5; 1; 1; 2; 2; 3; 4; 5; 5];

start_val = 1;
end_val = 5;

cycle_end_idx = [0; find(diff(row) == start_val-end_val); numel(row)]

cycle_end_idx = 4×1

0 5 12 20

n_cycles = numel(cycle_end_idx)-1;
cycle_length = diff(cycle_end_idx);
M = zeros(n_cycles,max(cycle_length));
for ii = 1:n_cycles
    M(ii,1:cycle_length(ii)) = row(cycle_end_idx(ii)+1:cycle_end_idx(ii+1));
end
disp(M);

     1     2     3     4     5     0     0     0
     1     1     2     3     4     5     5     0
     1     1     2     2     3     4     5     5

Ultimately, however, you may find that it's better to store those sub-sequences in a cell array rather than a matrix with extra zeros:

C = cell(1,n_cycles);
for ii = 1:n_cycles
    C{ii} = row(cycle_end_idx(ii)+1:cycle_end_idx(ii+1)).';
end
format compact
celldisp(C);

C{1} =
     1     2     3     4     5
C{2} =
     1     1     2     3     4     5     5
C{3} =
     1     1     2     2     3     4     5     5

[HARINI R]

Thanks.

What change i can do to the above code if the starting value for the sub-sequences is not always 1 but 2 in some cases and the ending is fixed at 5?

row = [1; 2; 3; 4; 5; 2; 3; 4; 5; 5; 2; 2; 3; 4; 5; 5];

[Voss]

Here's one way:

row = [1; 2; 3; 4; 5; 2; 3; 4; 5; 5; 2; 2; 3; 4; 5; 5];

start_val = 2;
end_val = 5;

cycle_end_idx = [0; find(diff(row) <= start_val-end_val); numel(row)];

n_cycles = numel(cycle_end_idx)-1;
cycle_length = diff(cycle_end_idx);
M = zeros(n_cycles,max(cycle_length));
for ii = 1:n_cycles
    M(ii,1:cycle_length(ii)) = row(cycle_end_idx(ii)+1:cycle_end_idx(ii+1));
end
disp(M);

     1     2     3     4     5     0
     2     3     4     5     5     0
     2     2     3     4     5     5

And here's a more general method that handles any sequence of non-decreasing cycles like these - so that you can detect the end of a cycle by seeing where the difference between an element and the previous element is negative. (The only difference between the three methods I've shown is how cycle_end_idx is defined.)

row = [1; 2; 3; 4; 5; 2; 3; 4; 5; 5; 4; 5; 6; 7];

cycle_end_idx = [0; find(diff(row) < 0); numel(row)];

n_cycles = numel(cycle_end_idx)-1;
cycle_length = diff(cycle_end_idx);
M = zeros(n_cycles,max(cycle_length));
for ii = 1:n_cycles
    M(ii,1:cycle_length(ii)) = row(cycle_end_idx(ii)+1:cycle_end_idx(ii+1));
end
disp(M);

     1     2     3     4     5
     2     3     4     5     5
     4     5     6     7     0

[HARINI R]

Very Helpful. Thanks a lot.

[Voss]

You're welcome!

[HARINI R]

Assuming its a sequence of non decreasing cycle starting from 0 and ending with 5,

but each cycle has different length...like the following:

row = [0;1;1;2;2;3;4;5;0;1;2;3;4;5;5;0;1;1;2;4;5;0;1;2;3;4;5]

after executing the above code, the output will be,

M =

0 1 1 2 2 3 4 5

0 1 2 3 4 5 5 0

0 1 1 2 4 5 0 0

0 1 2 3 4 5 0 0

But, Instead of padding the last missing column with zeros, how can i expand or shrink the cycle to my desired number of samples (for eg: 6) per cycle, so that i always have six columns no matter how big or small the cycle length is.

Thanks.

[Voss]

What should be the output for row = [0;1;1;2;2;3;4;5;0;1;2;3;4;5;5;0;1;1;2;4;5;0;1;2;3;4;5] with imposed cycle length (i.e., desired number of columns in output matrix) of 5?