Here is a developed example:
>> data = [1 2 3 4 -2 -3 4 6 6 -3 -6 -5 -7 2 1 5] ;
We try to develop a solution that detects the location of transitions between, groups of positive (incl. 0) numbers and groups of negative numbers.
>> gteqZero = data >= 0
gteqZero =
1 1 1 1 0 0 1 1 1 0 0 0 0 1 1 1
Places were there is a transition indicate groups boundaries; transitions are indicated by a non-zeros difference between consecutive elements:
>> diff(gteqZero)
ans =
0 0 0 -1 0 1 0 0 -1 0 0 0 1 0 0
As you can see, this could be used to spot ends of all groups but the last. We can add two non-zero entries to build a vector suitable for spotting starts of all groups, including the one that would come after the last..
>> [1, diff(gteqZero), 1]
ans =
1 0 0 0 -1 0 1 0 0 -1 0 0 0 1 0 0 1
>> groupStart = find([1, diff(gteqZero), 1])
groupStart =
1 5 7 10 14 17
and you can check that groups start at these positions. We included the starting position of a non-existent group that would start after the last group so we can compute the size of each group, as the difference between starting positions:
>> groupSize = diff(groupStart)
groupSize =
4 2 3 4 3
Now we have quite a few vectors that we can use to group your data. Which one(s) we use depends on what you want to achieve. Assuming that you want to create a cell array of groups, we can use group sizes to indicate MAT2CELL where to cut data:
>> group = mat2cell(data, 1, groupSize)
group =
[1x4 double] [1x2 double] [1x3 double] [1x4 double] [1x3 double]
>> group{1}
ans =
1 2 3 4
>> group{2}
ans =
-2 -3
etc..
Let me know if you have any question. Once you fully understand this example, let me know if you want to compute some statistics per group instead of splitting data into a cell array of groups.
EDIT: note the the whole can be aggregated into
>> groupSize = diff(find([1, diff(data>=0), 1])) ;
>> group = mat2cell(data, 1, groupSize)
