3 views (last 30 days)
T4H14 on 10 Oct 2017
Answered: Guillaume on 11 Oct 2017
How do I create a matrix A which is equal to a matrix X with column vectors added onto it which are the element wise square and cube of some of the columns of X? I have a matrix that is 32413x46 and I want to take columns 12-18, square them and add them as columns 47-53 to the previous matrix and then take those same columns 12-18, cube them and add them on as columns 54-60.
jean claude on 10 Oct 2017
can you give a numerical example ?

Guillaume on 11 Oct 2017
Right, you don't have a matrix but a table. Using correct term avoids wasting everyone time. matrices and tables are two completely different things.
All columns of a table must be named. I question the wisdom of adding so many columns to a table, particularly as the value are trivial to calculate whenever you need them, but if it needs to be done, the simplest thing would be to prepend (or append) the names of the existing columns with 'squared' and 'cubed' respectively. This is easily done:
yourtable = array2table(randi([0 100], 20, 46)); %demo table
newtable = [yourtable, ...
array2table(yourtable{:, 12:18} .^2, 'VariableNames', compose('squared_%s', string(yourtable.Properties.VariableNames(12:18)))), ...
array2table(yourtable{:, 12:18} .^3, 'VariableNames', compose('cubed_%s', string(yourtable.Properties.VariableNames(12:18))))]

Guillaume on 10 Oct 2017
If I understood correctly:
A = X;
A(:, 47:53) = X(:, 47:53) + X(:, 12:18) .^ 2;
A(:, 54:60) = X(:, 54:60) + X(:, 12:18) .^ 3;

Matthew on 10 Oct 2017
%I have a matrix that is 32413x46
X = rand(32413,46);
%and I want to take columns 12-18, square them and add them as columns 47-53 to the previous matrix
X(:,47:53) = X(:,12:18).^2;
%and then take those same columns 12-18, cube them and add them on as columns 54-60.
X(:,54:60) = X(:,12:18).^3;
Another possibility - if you mean "add" literally (as Guillaume answered above)
%I have a matrix that is 32413x46
X = rand(32413,46);
%and I want to take columns 12-18, square them and add them as columns 47-53 to the previous matrix
X(:,47:53) = X(:,12:18).^2 + X(47:53);
%and then take those same columns 12-18, cube them and add them on as columns 54-60.
X(:,54:60) = X(:,12:18).^3 + X(54:60);
T4H14 on 10 Oct 2017
You are right. I meant concatenate rather than add.
T4H14 on 10 Oct 2017
Unfortunately I actually have a 32413x46 Table rather than a Double so it won't let me perform that operation. Can I instead individually concatenate the square of column variables? Do I have to then name these new columns too?