Generate Variable Names with a Loop

I have a simple question that I've been unable to find an answer for. I'd like to make n variable of the format:

variable_1 variable_2 variable_... variable_n

I know I should be using a for loop, but I cant figure out how to concatenate the string with the counter.

Thanks

[Stephen23]

https://www.mathworks.com/help/matlab/matlab_prog/string-evaluation.html

https://www.mathworks.com/matlabcentral/answers/304528-tutorial-why-variables-should-not-be-named-dynamically-eval

Answer 1

[David Sanchez]

N = 10;          % number of variables
%method 1
for k=1:N
    temp_var = strcat( 'variable_',num2str(k) );
    eval(sprintf('%s = %g',temp_var,k*2));
end

% method 2, more advisable
for k=1:N
    my_field = strcat('v',num2str(k));
    variable.(my_field) = k*2;
end

[john]

Thanks, num2str works perfectly.

[Jan]

While the first method using eval is a bad idea, the 2nd can be abbreviated a little bit:

    my_field = sprintf('v%d', k);

[Stephen23]

Forcing meta-data into variable names is slow and complex.

Indexing is much simpler and much more efficient:

% method 3, even more advisable:
C = cell(1,N);
for k = 1:N
    C{k} = k*2;
end

[Dav]

Thanks for this method Stephen. Short, simple, and works perfectly.

[Soumya Mishra]

Hi. I tried using this in my program, I was expecting a 1X100 cell array and each cell value must show 23x20 double. But unfortunately all my cells show blank []

[Soumya Mishra]

clc
clear all
close all
% x = [];
% y = [];
t0 = readtable('master_mfcc_positive.xlsx');
t1 = readtable('mfcc_master_negative.xlsx');
S0 = table2array(t0);
S1 = table2array(t1);
% S = vertcat (S0, S1);
  m1=12; 
m2=11;
p = 100;
for i = 1:m1:1200
    for j = 1:m2:1100
            
for k = 1:p
    C = cell(1,p);
    C{k} = vertcat(S0(i:k*m1,:), S1(j:k*m2,:));
% y{p} = x{p}((i:p*m1 + j:p*m2),19);
        end
    end
end

[Walter Roberson]

for k = 1:p
    C = cell(1,p);
    C{k} = vertcat(S0(i:k*m1,:), S1(j:k*m2,:));
% y{p} = x{p}((i:p*m1 + j:p*m2),19);
end

Every iteration of k, you throw away whatever was in C and replace it with a cell array containing empty entries. Then you assign something into one of those entries.

By the time you reach the end of the k loop, only the last of the entries will have anything stored in it.

The last iteration of i will be i = 1189 . The last iteration of j will be j = 1090. So you will extract S0(1189:100*12,:)) and S1(1090:100*11, :) and concatenate them together. That is 12 rows and 11 rows, total of 23 rows, which checks out with your requirement for 23 rows. If only you were not

C = cell(1,p)

each iteration of k. And if only you used C after the end of the for k loop so that you did not effectively throw away the work for all of the i and j iterations except the last.

[Soumya Mishra]

Thanks a lot Walter. I understood your point. But please help me with getting something like C{1}, C{2}....C{100}

I intend to go for a classification problem and i have to use 100 pairs of C(for X) and Y for creating SVM Model.

My aim was to use a loop for getting :

% x2 = vertcat(S1, S10);
% y2 = x2(:,20);
% x3 = vertcat(S2, S11);
% y3 = x3(:,20);
% x4 = vertcat(S3, S12);
% y4 = x4(:,20);
% x5 = vertcat(S4, S13);
% y5 = x5(:,20);
% x6 = vertcat(S5, S14);
% y6 = x6(:,20);
% x7 = vertcat(S6, S15);
% y7 = x7(:,20);
% x8 = vertcat(S7, S16);
% y8 = x8(:,20);
% x9 = vertcat(S8, S17);
% y9 = x9(:,20);
% x10 = vertcat(S8a, S17a);
% y10 = x10(:,20);
% x11 = vertcat(S8b, S17b);
% y11 = x11(:,20);
% 
% x12 = vertcat(S8c, S17c);
% y12 = x12(:,20);
% 
% x13 = vertcat(S8d, S17d);
% y13 = x13(:,20);
% 
% x14 = vertcat(S8e, S17e);
% y14 = x14(:,20);
% 
% x15 = vertcat(S8f, S17f);
% y15 = x15(:,20);
% 
% % score = LDA(x7, y7)
% % [index,featureScore] = feature_rank(x7,y7,15,'Fisher_Score')
% SVMModel1 = fitcsvm(x1,y1,'Standardize',true,'KernelFunction','RBF',...
%      'KernelScale','auto');
%  SVMModel2 = fitcsvm(x2,y2,'Standardize',true,'KernelFunction','RBF',...
%      'KernelScale','auto');
%  SVMModel3 = fitcsvm(x3,y3,'Standardize',true,'KernelFunction','RBF',...
%      'KernelScale','auto');
%  SVMModel4 = fitcsvm(x4,y4,'Standardize',true,'KernelFunction','RBF',...
%      'KernelScale','auto');
%  SVMModel5 = fitcsvm(x5,y5,'Standardize',true,'KernelFunction','RBF',...
%      'KernelScale','auto');
%  SVMModel6 = fitcsvm(x6,y6,'Standardize',true,'KernelFunction','RBF',...
%      'KernelScale','auto');
%  SVMModel7 = fitcsvm(x7,y7,'Standardize',true,'KernelFunction','RBF',...
%      'KernelScale','auto');
%  SVMModel8 = fitcsvm(x8,y8,'Standardize',true,'KernelFunction','RBF',...
%      'KernelScale','auto');
%   SVMModel9 = fitcsvm(x9,y9,'Standardize',true,'KernelFunction','RBF',...
%      'KernelScale','auto');
%  SVMModel10 = fitcsvm(x10,y10,'Standardize',true,'KernelFunction','RBF',...
%      'KernelScale','auto');
%  SVMModel11 = fitcsvm(x11,y11,'Standardize',true,'KernelFunction','RBF',...
%      'KernelScale','auto');

[Walter Roberson]

Why are you creating numbered variables?

http://www.mathworks.com/matlabcentral/answers/304528-tutorial-why-variables-should-not-be-named-dynamically-eval

[Florian Flaig]

Hello everybody,

I use variant 1, because I want the name of the structure to have the sequential number.

Now I want to store matrices in this structure.

When I use the eval(sprintf variant, it stores only the first value of the matrix under the name and not the whole matrix.

A=[3,5]

% B is a 3rd order tensor.
% => B(:,:,3) is a matrix
% => B(:,:,5) is another matrix

for k = 1:2
    temp_var = strcat( 'variable_',num2str(A(i)),'.matrix' );
    uTemp = B(:,:,A(i));
    eval(sprintf('%s = %g',temp_var,uTemp));
end

I expect 2 matrices.

variable_3.matrix % and
variable_5.matrix

instead only the value from the top left is stored there.

What do I have to do differently?

Greetings and thank you,

Florian

[Stephen23]

"What do I have to do differently?"

1. fix your loop iterator: the loop iterates over k, but inside the loop you refer only to i (which is undefined in your code, but presumably is defined in the workspace, thus also illustrating why experienced MATLAB users avoid scripts for reliable code).
2. https://www.mathworks.com/matlabcentral/answers/304528-tutorial-why-variables-should-not-be-named-dynamically-eval (something else that experienced MATLAB users avoid)

Answer 2

[Phil Kühnholz]

Hi im using the 2nd method for a projekt im working on, in short i used the variables to mark images

 my_field = strcat('v',num2str(k));
                img = im2double(img);
                variable.(my_field) = img;

I have a bunch of subplots that are filled red, if they are clicked on i want the fitting picture to be displayed.

it works if a put in the variable manually

a=click; b=str2num(cell2mat(a))
subplot(n,n,b(1));
imshow(variable.v1)

but it only gives out black blocks, nothing, or dosnt even work if i try to construct the variables name

a=click; b=str2num(cell2mat(a));
%none of these work
varistr = strcat('variable.v',num2str(b(1)));
varid = str2num(varistr1);
subplot(n,n,b(1));
imshow(varid);

%or
varid = [str2double('variable.v'),b(1)]
subplot(n,n,b(1));
imshow(varid)

%or 
subplot(n,n,b(1));
imshow(variable(b(1)))

Any idea on how i can fix this?

[Stephen23]

Using indexing would be much simpler.

But if you insist on this indirect approach, you probably just need to generate the appropriate fieldname:

https://www.mathworks.com/help/matlab/matlab_prog/generate-field-names-from-variables.html

f=sprintf('v%d',a{1});
variable.(f)

[Phil Kühnholz]

works fantastic ^^ thank u so much.