Adapting a dataset such that it contains unique values

Hello,

I'm facing the following problem:

I have a matrix A(X,1) which contains time data. The time is given in steps of 20 seconds, so 0, 30, 75 etc. means for example 05:00:00, 05:10:00, 05:25:00 [hrs:min:sec] respectively.

Now the problem is some actions X have the same time value. This shouldn't be the case as each action should take place at a unique time.

What I want to do is to add a timestep of 20 or 40 seconds to a time (i) which is the same as the value before (i-1). Though the new value (i) shouldn't overshoot or equal the time of the next value (i+1).

An example is the following:

A(1:6,1)= 225   [06:15:00]
A(7:19,1)= 240  [06:20:00]
A(20:22)= 255   [06:25:00]

Now for the first case A(1:5,1) there are only 6 values which are the same. As a result, the timevalues of A(2:6,1) should be adapted such that they are not equal to A(1,1) and are still smaller than A(7,1). To do so, 2 time steps can be added to each value. I.e. A(2,1)= 227, A(3,1)= 229 etc.

However, for A(7:19,1) only 1 time step can be added as 2 time steps will lead to an overshoot w.r.t. A(20,1).

So if there are <8 of the same time values, 2 time steps can be added; if >=8 then only 1 time step can be added.

Can anyone help me writing a code for this procedure?

Answer 1

[Guillaume]

Possibly,

A = [repmat(225,6,1); repmat(240,13, 1); repmat(255,3, 1)]
startidentical = find(diff([-Inf; A; Inf])); %location of start of each sequence, and 1 past the end
A(end+1) = A(end) + diff(startidentical(end-1:end)) * 2;  %add extra time for last sequence
for seqiter = 1:numel(startidentical)-1
    seqindices = startidentical(seqiter) : startidentical(seqiter+1)-1;
    A(seqindices) = A(seqindices) + (0:numel(seqindices)-1).' * floor((A(startidentical(seqiter+1)) -  A(startidentical(seqiter))) / numel(seqindices));
end
A(end) = []  %remove added element

[Bas]

Thanks that helped :)