MatLab Group by Non-Descending Order - matlab

Question

In MatLab 2016b, is there a way to group the values of the below vector by non-descending order?
values = [1 1 1 7 17 74 89 91 96 1 5 32 43 78 84 95 98 100 0 0 15 31 69 88 94 97 100 100 100 0 2 12 42 66 78 83 89 94 97 1 6 34 63 65 75 89 93 98]
so I will get the following groups:
group A) 1 1 1 7 17 74 89 91 96
group B) 1 5 32 43 78 84 95 98 100
group C) 0 0 15 31 69 88 94 97 100 100 100
group D) 0 2 12 42 66 78 83 89 94 97
group E) 1 6 34 63 65 75 89 93 98

Answer 1

The following code should give you what you are looking for:
% Define the values...
values = [1 1 1 7 17 74 89 91 96 1 5 32 43 78 84 95 98 100 0 0 15 31 69 88 94 97 100 100 100 0 2 12 42 66 78 83 89 94 97 1 6 34 63 65 75 89 93 98];
% Define the linear indices...
subs = 1:numel(values);
% Define the grouping indices...
idx = [0 subs(diff(values) < 0) subs(end)];
% Split the values into multiple arrays...
result = arrayfun(#(k)values(idx(k-1)+1:idx(k)),2:numel(idx),'UniformOutput',false);
Here is the final output:
>> result{:}
ans =
1 1 1 7 17 74 89 91 96
ans =
1 5 32 43 78 84 95 98 100
ans =
0 0 15 31 69 88 94 97 100 100 100
ans =
0 2 12 42 66 78 83 89 94 97
ans =
1 6 34 63 65 75 89 93 98

Answer 2

diff(values)
returns a vector with the difference between each element of values and the following element. We can use this in a comparison to give a logical vector:
diff(values) < 0
returns a vector containing true, or 1, whenever the corresponding element of values is less than the following element. To finish the job we can use the cumsum function which returns a vector containing the running total of the elements in the input vector:
groups = cumsum([1, diff(values) < 0])
Just one wrinkle: we have to prepend an extra element at the front of the diff vector, because it has one fewer element than values did. Now we have a vector groups containing 1 for each element of values that should be in the first group, 2 for each element that should be in the second group, and so on.
You can access the individual groups using logical indexing:
values(groups == n) % returns a vector of the values in group n
If you want the groups as separate vectors you could use arrayfun:
groupArray = arrayfun(#(x) values(groups == x), 1:max(groups), 'UniformOutput', false)
This returns a cell array where groupArray{1} is the first group, groupArray{2} is the second group, etc.