One possible way to do it:
data = [1 1 1 53.1100
1 2 2 59.0900
2 1 2 71.3100
2 2 1 70.2400];
%first split data into two matrices for easier referencing:
rowdest = data(:, 1:end-1);
rowvals = data(:, end);
%second create matrix of column indices
coldest = repmat(1:size(rowdest, 2), size(rowdest, 1), 1);
%third use accumarray to compute the sum:
newdata = accumarray([rowdest(:) coldest(:)], repmat(rowvals, size(rowdest, 2), 1))
