Matrix operations that preserve floating point precision

When concatenating a matrix from the row element of a cell array, I find that the floating points are always converted to integers. How do avoid this?

E.g.

 Q = {[1;2;3],[4.000;5.000;6.000],[7e+04;8e+03;9e+02],[0.1234;0.34;0.4]};
 P = [Q{1}];
 for i=2:4
    P = cat(2,P,Q{i});
 end

Answer 1

[Sean de Wolski]

The values will be converted to the most restrictive class so preconvert the integers to double:

Q = cellfun(@double,Q,'UniformOutput',0);

Also rather than the for-loop you can use the comma-separated list expansion of the cell array:

P = horzcat(Q{:});

More Details

Of course if you would like the best of both worlds (floats and integers) and would like to have them in a non-cell array (and you have the Statistics Toolbox) then you could use a dataset

Q = {uint8([1;2;3]),uint32([4.000;5.000;6.000]),[7e+04;8e+03;9e+02],[0.1234;0.34;0.4]};
D = dataset(Q{:})

[Tolulope]

Yep I have got the Stat Toolkit, so the dataset function works a treat. Thanks