MATLAB: Find the median of sequential duplicate entries

referenceunique

Hi! So I have a matrix
 num =
     1     4
     2     4
     3     4
     4     4
     5     5
     6     6
     7     6
     8     6
     9     6
    10     6
    11     7
   ....
    31    14
    32    14
    33    13
    34    12
    35    12
    36    12
    37    12
    38    12
    39     9
    40     9
    41     9
    42     9
    43     8
    44     8
    45     8
    46     8
    47     5
    48     5
    49     4
    50     3
And I want to take the median of the x entries with sequential y entries that are duplicate such that
inum =
     2.5   4
     5     5
     8     6
    11     7
    ....
    31.5  14
    33    13
    36    12
    40.5   9
    44.5   8
    47.5   5
    49     4
    50     3
I am using unique such that
[C,ia,ic] = unique(num(:,2),'stable')
OutMatrix = [accumarray(ic, num(:,1), [], @median ) ,C];
However, my outmatrix is showing the median of all the x entries that have y values that are the same i.e It took the median of 1,2,3,4,49 and not the median on 1,2,3,4 and 49 How do I do this? Thanks

Best Answer

One way to achieve this is to replace num(:,2) with a monotonically increasing group/block ID that you build. Here is how:
 blockStart = [true; diff(num(:,2)) ~= 0] ;
 blockId    = cumsum( blockStart ) ;
applied to the numbers that you gave, this builds the 3rd column below;
 >> [num, blockId]
 ans =
     1     4     1
     2     4     1
     3     4     1
     4     4     1
     5     5     2
     6     6     3
     7     6     3
     8     6     3
     9     6     3
    10     6     3
    11     7     4
    31    14     5
    32    14     5
    33    13     6
    34    12     7
    35    12     7
    36    12     7
    37    12     7
    38    12     7
    39     9     8
    40     9     8
    41     9     8
    42     9     8
    43     8     9
    44     8     9
    45     8     9
    46     8     9
    47     5    10
    48     5    10
    49     4    11
    50     3    12
and you can see that, using this, you eliminate the interference between blocks with same num(:,2) on both sides of the peak. Then you perform the call to ACCUMARRAY that you already implemented, using blockId this time:
 >> OutMatrix  = [accumarray( blockId, num(:,1), [], @median ), ...
                  num(blockStart,2)]
 OutMatrix =
    2.5000    4.0000
    5.0000    5.0000
    8.0000    6.0000
   11.0000    7.0000
   31.5000   14.0000
   33.0000   13.0000
   36.0000   12.0000
   40.5000    9.0000
   44.5000    8.0000
   47.5000    5.0000
   49.0000    4.0000
   50.0000    3.0000
Related Question