Groupsummary on timetable with cell/matrix data

Hello,

Is it possible with groupsummary to apply aggregation function to cell or matrix data?

For example, if I want to average the "Matrix" every 1s with the following timetable example:

    t        Angle        Matrix    
    ________    _____    ______________

    0.2 sec     180.4    {53×53 double}
    0.4 sec     180.4    {53×53 double}
    0.6 sec     180.4    {53×53 double}
    0.8 sec     180.4    {53×53 double}
    1 sec       180.4    {53×53 double}
    1.2 sec     180.4    {53×53 double}
    1.4 sec     180.4    {53×53 double}
    1.6 sec     180.4    {53×53 double}
    1.8 sec     180.4    {53×53 double}

Thank you

Gregory

[dpb]

Average the Matrix how, precisely? Overall, by column, row, element ...???

[Gregory Bolard]

Hello, I would like the overall matrix to be averaged every second actually.

[Image Analyst]

I don't see how groupsummary() applies. Which column is your group? It looks like every row in your table is a distinct group. For simplicity I'd just do (untested)

for k = 1 : size(yourTable, 1)
    thisMatrix = table{k, 3};
    means(k) = mean(thisMatrix(:));
end

[dpb]

It's the accumulation mode over minutes I think OP wants, IA. See below for what I think is intended.

I had to go dig in the doc meself; I didn't recall that facility being built into it.

Answer 1

[dpb]

Took a little head-scratching, but the following appears to work as desired--

fnMN3=@(c){mean(cat(3,c{:}),3)};        % function to catenate 2D cell array to 3D array, take mean
means=groupsummary(T,'t','minute',fnMN3,'Matrix');

with a sample table here

>> groupsummary(T,'t','minute',fnMN3,'M');
ans =
  1×3 table
       minute_t        GroupCount       fun1_M   
    _______________    __________    ____________
    [0 sec, 60 sec]        6         {5×5 double}
>>

that matched for a set of random integers had created to make a dummy table.

t=seconds(0:0.2:1);
M=randi(100,5,5,numel(t));
T=[array2table(t.','VariableNames',{'t'}) vertcat(cell2table(M(:),'VariableNames',{'M'}))]
T =
  6×2 table
       t            M      
    _______    ____________
      0 sec    {5×5 double}
    0.2 sec    {5×5 double}
    0.4 sec    {5×5 double}
    0.6 sec    {5×5 double}
    0.8 sec    {5×5 double}
      1 sec    {5×5 double}
>> T.M{:}
ans =
    31    73    81    86    96
    89    71    88    18    42
    16    47    46    65    29
    62    60    23    68    50
    89    19    29    37    81
ans =
    29    20    61    50     4
     1    49    81    27    55
     5     3    30    76    23
    11    63    18    30    38
    85    11    45     8    30
ans =
    74    46    51    97     7
    15    25    16    31    31
     8     2    59    46    33
    94    38    22    91    21
    92    65    15    45    18
ans =
    73    62    27    84    28
    19    22    10    58    22
    63    73    94    65    32
    29    37    61    50    31
    78    23    31    19    89
ans =
    65    77    76    72     2
    32    33    26    12    22
    83    29   100    39    73
    29    67    78    92    68
    47     3    71    30    52
ans =
    19    24     5    97    39
    47    32    45    94    24
    27    42    77    17    87
    49    23    35    55    26
    46    54    24    91    83

>> mean(cat(3,T.M{:}),3)
ans =
   48.5000   50.3333   50.1667   81.0000   29.3333
   33.8333   38.6667   44.3333   40.0000   32.6667
   33.6667   32.6667   67.6667   51.3333   46.1667
   45.6667   48.0000   39.5000   64.3333   39.0000
   72.8333   29.1667   35.8333   38.3333   58.8333
>>

[Gregory Bolard]

Thank you very much, this solves my problem!

Best regards,

Gregory