How to take the average of the columns of a matrix and insert into a column matrix

I have a piece of code that reads a .txt file with several matrices and sorts them into cell arrays. Each of the matrices has always three columns with varying rows. My question would be how i can take the average of the first two columns column from these matrices separately and put it in a column matrix that would always start the row with 0 and then the average of the first column, followed by the average of the second column and then a value that i would insert with a prompt or something. One more thing is that the last value of this new column matrix is also something that i would insert. Here's the code for the code and the .txt file is indexed.

filepath = '/Users/...';
fileID = fopen(filepath);
nGroups = 0;
stationGroups = cell(0,1);
while true
    nStations = fgets(fileID);
    if ~ischar(nStations)
        break;
    end
    nGroups = nGroups + 1;
    nStations = str2num(nStations);
    stationMatrix = zeros(nStations,3);
    for currentStation = 1:nStations
        fprintf('Loading station %d of station group %d: ', ...
            currentStation, nGroups);
        testStation = fgets(fileID);
        if ~ischar(testStation)
            error('Error - file ended too early!');
        end
        testStation = str2num(testStation);
        if numel(testStation)~=3
            error('Error - Station %d didn''t have three values!', ...
                currentStation);
        end
        fprintf('%.2f ',testStation);
        fprintf('\n');
        stationMatrix(currentStation,:) = testStation;
    end
    stationGroups{end+1} = stationMatrix;
    nGroups = length(stationGroups);
end
fclose(fileID);

The format of the column matrix would then be(for if i choose only the first two matrices of the file):

0
average of the first column
average of the second column
value that i insert
0
average of the first column
average of the second column
value that i insert 
another value that i insert

practically it would look like this:

                     0
     0.397309273000000
     9.579781409000001
    -0.008595000000000
                     0
     0.397309273000000
     9.579781409000001
    -0.008595000000000
     0.005900000000000

Answer 1

[Star Strider]

A slight variation on my previous code works here:

fidi = fopen('file.txt','rt');
c = textscan(fidi, '%f%f%f', 'Delimiter',' ', 'CollectOutput',1);
fclose(fidi);
d = [c{:}];                                                             % Can Use ‘cell2mat’ Here As Well
bk = any(isnan(d), 2);                                                  % Rows With ‘any’ NaN Columns
nr = d(bk,1);                                                           % Number Of Rows In Next Segment
de = d(~bk,:);                                                          % Eliminate Rows With NaN Columns
mtx = mat2cell(de, nr, 3);                                              % Use Those Data To Create The ‘mtx’ Matrix
OutC = cellfun(@(x)[0 mean(x(:,[1 2])) NaN]', mtx, 'Uni',0);            % Cell Array Of Mean Values
OutM = cell2mat(OutC);                                                  % Double Array Of Mean Values

OutM =  
            0
     3.97e+05
   9.5811e+06
          NaN
            0
   3.9831e+05
   9.5775e+06
          NaN
       ...

Put your other inserted values in the NaN positions. The easiest way would be to create them in a (nnz(bk)x1) vector, and then assign them to the NaN values in the ‘OutM’ vector:

Added = rand(nnz(bk),1);                                                % Use Random Values To Test
OutM(isnan(OutM)) = Added;

[Feliciano Döring]

Thanks again! Just one more thing, how would i add the last value? just assigning OutM to a variable like m=[OutM;value];?

[Star Strider]

As always, my pleasure!

The easiest way to add the last value would be to create a vector for all of them that is nnz(isnan(OutM)) long, (that is, with dimension (nnz(isnan(OutM)) x 1)). Then write it to ‘OutM’ as I illustrated with the ‘Added’ vector earlier.

Otherwise, the indices of the NaN values are:

NaNidx = find(isnan(OutM));

and you could use that to write to any of them individually, so if you wanted to write ‘m’ to NaN #3:

OutM(NaNidx(3)) = m;

I tested that to be certain. It works.