How to use command function?

Hi everyone,

I have to generate 15 matrices of dimensions 432x6. Instead of doing it manually, such as:

%create a matrix with the forecasted values of all currencies for the 1 predictor.
for i=1:6;
    for t=1:432;
predictor_1(t,i)=k1_total(t,1+15*(i-1));
    end;
end;
%create a matrix with the forecasted values of all currencies for the 2 predictor.
for i=1:6;
    for t=1:432;
predictor_2(t,i)=k1_total(t,2+15*(i-1));
    end;
end;
%create a matrix with the forecasted values of all currencies for the 3 predictor.
for i=1:6;
    for t=1:432;
predictor_3(t,i)=k1_total(t,3+15*(i-1));
    end;
end;

and so on. k1_total is a 432x90 matrix. How could I use the ''function'' command in order to avoid all these replications and look more coherent.

I 've tried the following:

for i=1:size(k1_total,2);
   [predictor_i,xxx,xxx,xxx,xxx,xxx,xxx]=Create_predictors(...
      k1_total(:,i));
 results_ECON(i,:)=[predictor_i];
end;

and on the script, I use:

function [predictor_i,SR,weight_risky,cumulative_return,avg_turnover,Sortino,G]=...
Create_predictors(k1_total,ir_i,target_volatility,actual_risky_return_i,ir_us_lag,k1_uk)

for i=1:6;
    for t=432;
predictor_k_econ(t,i)=k1_total(t,1+15*(i-1));
    end;
end;

But I get an error of mismatching dimensions. I would appreciate any help. Thanks in advance.

Answer 1

[James Tursa]

I would advise using a cell array for this instead of many separately named variables. E.g.,

predictor = cell(15,1);
for k=1:15
    predictor{k} = k1_total(:,k:15:end);
end

So predictor{1} would be used downstream instead of your predictor_1, etc. This has the advantage of using indexing downstream in your code instead of hard-coded named variables.