How can I save my results of a loop in one table

I have 25 parameter combination in my code (b&T --> 1:5).

In the third part of my code I am doing the anderson darling test on my combinations.

It tells me if ONE combination ( for example b=1 and T=1) is a good combination or bad. I want to see the results for all 25 combination. But with this code, I only see the last combination of the loop.

How can I save them all in a table?

clear all;
n   = 10;
t0  = 0.5;
b = 1:5;
T = 1:5;

for v_T= 1:length(T)
    for v_b= 1:length(b)
        data(:,v_b,v_T) = wblrnd(v_b,v_T, [n,1]) + t0;
        start       = [1 0 0];
        custompdf   = @(x,a,b,c) (x>c).*(b/(a-c)).*(((x-c)/(a-c)).^(b-1)).*exp(-((x-c)/(a-c)).^b);
        opt         = statset('MaxIter',1e3,'MaxFunEvals',1e3,'FunValCheck','off');
        params(v_b,1:3,v_T) = mle(data(:,v_b,v_T),'pdf',custompdf,'start',start,'Options',opt,'LowerBound',[0 0 0],'UpperBound',[Inf Inf Inf])
    end
end

for v_T= 1:length(T)
  for v_b= 1:length(b)
 T_A = T(v_T)
 b_A = b(v_b)
dist = makedist('weibull','a',T_A,'b',b_A)
[h,p] = adtest(data(:, v_b, v_T),'Distribution',dist)
  end 
 end

Answer 1

[Star Strider]

The adtest function returns scalars for the outputs, so something like this could work:

[h(v_b, v_T),p(v_b, v_T)] = adtest(data(:, v_b, v_T),'Distribution',dist);

That will save the results in their respective matrices.

I cannot run your code, so I cannot test that.

[Mustafa Vural]

It helped me a lot, thank you very much.

[Star Strider]

As always, my pleasure!

If you actually want to save them in a table object, try this:

[vb,vT] = ndgrid(1:length(T),1:length(b));
T1 = table(vT(:), vb(:), h(:), p(:), 'VariableNames',{'v_T','v_b','h','p'});

.

[Mustafa Vural]

It displays:

Error using table (line 233)

All table variables must have the same number of rows.

for v_T= 1:length(T)
  for v_b= 1:length(b)
 T_A = T(v_T)
 b_A = b(v_b)
dist = makedist('weibull','a',T_A,'b',b_A)
[h(v_b, v_T),p(v_b, v_T)] = adtest(data2p(:, v_b, v_T),'Distribution',dist)
[vb,vT] = ndgrid(1:length(T),1:length(b )); 
T1 = table(vT(:), vb(:), h(:), p(:), 'VariableNames',{'v_T','v_b','h','p '}); 

  end 
end

[Star Strider]

You need to put the ndgrid and table calls after the end of both loops. Then it works.

for v_T= 1:length(T)
    for v_b= 1:length(b)
        T_A = T(v_T)
        b_A = b(v_b)
        dist = makedist('weibull','a',T_A,'b',b_A)
        [h(v_b, v_T),p(v_b, v_T)] = adtest(data(:, v_b, v_T),'Distribution',dist)
    end
end

[vb,vT] = ndgrid(1:length(T),1:length(b));
T1 = table(vT(:), vb(:), h(:), p(:), 'VariableNames',{'v_T','v_b','h','p'});

.

[Mustafa Vural]

oh wooow it works so nice. Thats so cool! Thank you I appreciate that!

How can I do the same at the second part of my code? params are my estimated results.

First column are the estimated "T" values, second estimated "b" values and third estimated "t0" values.

It looks like this:

clear all;
n   = 10;
t0  = 0.5;
b = 1:5;
T = 1:5;
for v_T= 1:length(T)
    for v_b= 1:length(b)
        data(:,v_b,v_T) = wblrnd(v_b,v_T, [n,1]) + t0;
        start       = [1 0 0];
        custompdf   = @(x,a,b,c) (x>c).*(b/(a-c)).*(((x-c)/(a-c)).^(b-1)).*exp(-((x-c)/(a-c)).^b);
        opt         = statset('MaxIter',1e3,'MaxFunEvals',1e3,'FunValCheck','off');
        params(v_b,1:3,v_T) = mle(data(:,v_b,v_T),'pdf',custompdf,'start',start,'Options',opt,'LowerBound',[0 0 0],'UpperBound',[Inf Inf Inf])
        params3p(v_b,4,v_T)            = T_A;
        params3p(v_b,5,v_T)            = b_A; 
        params3p(v_b,6,v_T)            = t0;
        params3p(v_b,7,v_T)            = n;
    end
end

[Star Strider]

As always, my pleasure!

I am not able to run that.

The first problem is:

Unrecognized function or variable 'T_A'.
Error in ... 
        params3p(v_b,4,v_T)            = T_A;

I suspect there may be others like it.

[Mustafa Vural]

Ooh yes you right, I forgot something, now you can run it.

clear all;
n   = 10;
t0  = 0.5;
b = 1:5;
T = 1:5;
data3p              = zeros(n,length(b),length(T)); %Prelallocating
params3p            = zeros(length(b),7,length(T)); %Prelallocating

for v_T= 1:length(T)
    for v_b= 1:length(b)
 T_A = T(v_T)
 b_A = b(v_b)
        data3p(:,v_b,v_T) = wblrnd(v_b,v_T, [n,1]) + t0;
        start       = [1 0 0];
        custompdf   = @(x,a,b,c) (x>c).*(b/(a-c)).*(((x-c)/(a-c)).^(b-1)).*exp(-((x-c)/(a-c)).^b);
        opt         = statset('MaxIter',1e3,'MaxFunEvals',1e3,'FunValCheck','off');
        params3p(v_b,1:3,v_T) = mle(data3p(:,v_b,v_T),'pdf',custompdf,'start',start,'Options',opt,'LowerBound',[0 0 0],'UpperBound',[Inf Inf Inf])
        params3p(v_b,4,v_T)            = T_A;
        params3p(v_b,5,v_T)            = b_A; 
        params3p(v_b,6,v_T)            = t0;
        params3p(v_b,7,v_T)            = n;
    end
end

[Star Strider]

That was a bit more involved because of the extra index. I created the second index as ‘iv’ since I have no idea how your code would otherwise refer to it. (You can obviously name it anything you want.)

Try this:

[vT,iv,vb] = meshgrid(1:length(T), (1:7), 1:length(b));
T2 = table(vT(:), iv(:), vb(:), params3p(:), 'VariableNames',{'v_T','iv','v_b','params3p'});

with:

First_9_Rows = T2(1:9,:)

producing:

First_9_Rows =

  9×4 table

    v_T    iv    v_b    params3p
    ___    __    ___    ________

     1     1      1      1.8079 
     1     2      1      1.7048 
     1     3      1      2.1389 
     1     4      1      2.6878 
     1     5      1      3.7165 
     1     6      1     0.49299 
     1     7      1     0.72493 
     2     1      1      0.7776 
     2     2      1      0.5925 

Since ‘T2’ is a (175x4) tabble, and I do not understand what you are doing with this, I will let you check it to be certain it is correct. It might be necessasry to change the order of the meshgrid arguments, and the order of the associated outputs to match them, to represent the ‘params3p’ matrix correctly in ‘T2’.

[Mustafa Vural]

I am sry I couldnt explain it correctly. I appreciate your help and your time. Its not so complicated I explain it to you. But you really helped me a lot.

In total, I have 25 combination of my given parameter (b, T, and t0),

because b=1:5; T=1:5; t0=0,5 --> (t0 is just one number, so its included in every combination).

1. With wblrnd, I create random numbers and store it to data3p.
2. With mle, I estimate the random numbers and receive 25 combination of estimated values.
3. With anderson darling, I am testing if the random numbers in data3p are good results or bad.

• My first column are my 25 estimated T values.
• Second column: 25 estimated b values
• Third column: estimated t0 values
• Fourth column: given T values ( 1 1 1 1 1 2 2 2 2 2; and so on...)
• Fifth column: given b values ( 1 2 3 4 5 1 2 3 4 5; and so on...)
• Sixt column: given t0 values ( t0 t0 t0 t0 t0...)
• Seventh column: given n failure times ( n n n n n; ...)

If you run my below code, in "results3p", you can see how I store them. I thought I could change the result line, and do it like like you showed me in adtest, because it looks much more better. But you really helped me a lot, so I think can go on with this code.

clear all;
n   = 10;
t0  = 0.5;
b = 1:5;
T = 1:5;
data3p              = zeros(n,length(b),length(T)); %Preallocating
params3p            = zeros(length(b),7,length(T)); %Preallocating
results3p          = double.empty(length(b)*length(T), 0);

for v_T= 1:length(T)
    for v_b= 1:length(b)
 T_A = T(v_T)
 b_A = b(v_b)
        data3p(:,v_b,v_T) = wblrnd(v_b,v_T, [n,1]) + t0;
        start       = [1 0 0];
        custompdf   = @(x,a,b,c) (x>c).*(b/(a-c)).*(((x-c)/(a-c)).^(b-1)).*exp(-((x-c)/(a-c)).^b);
        opt         = statset('MaxIter',1e3,'MaxFunEvals',1e3,'FunValCheck','off');
        params3p(v_b,1:3,v_T) = mle(data3p(:,v_b,v_T),'pdf',custompdf,'start',start,'Options',opt,'LowerBound',[0 0 0],'UpperBound',[Inf Inf Inf])
        params3p(v_b,4,v_T)            = T_A;
        params3p(v_b,5,v_T)            = b_A; 
        params3p(v_b,6,v_T)            = t0;
        params3p(v_b,7,v_T)            = n;
    end
    results3p((v_T-1) * length(b) + 1:v_T*length(b), 1:size(params3p, 2))            = params3p(:,:,v_T);
end

[Star Strider]

I am sorry, however at this point I am lost.

Appparently, the various parts of ‘param3p’ are separate columns of the table, however I have no idea how to assign them different variable names, if they even need different variable names. So the result should be a (25x7) table?

Also, the ‘results3p’ variable is new. I have no idea where it fits with the rest of this, or if it even does.

[Mustafa Vural]

Yes, its a 25x7 in this case. If I change b to 1:3 and T to 1:3, its a 9x7 table.

params3p --> my estimatings for 3p weibull distribution.

params2p---> my estimatings for 2p weibull distribution.

This is my whole code:

clear all;
n   = 100;
t0  = 0.5;
b = 1:5;
T = 1:5;
rng('shuffle');

result2p            = double.empty(length(b)*length(T), 0);  % A = ClassName.empty(sz1,...,szN) returns an empty array with the specified dimensions. At least one of the dimensions must be 0.
data2p              = zeros(n,length(b),length(T)); % Preallocating: Geschwindigkeit des Durchlaufs erhöhen
params2p            = zeros(length(b),7,length(T)); 

result3p            = double.empty(length(b)*length(T), 0);
data3p              = zeros(n,length(b),length(T));
params3p            = zeros(length(b),7,length(T));

for v_T = 1:length(T)
   for v_b = 1:length(b)
 T_A = T(v_T)
 b_A = b(v_b)
 
        data2p(:,v_b,v_T)     = wblrnd(v_T,v_b, [n,1]);
        params2p(v_b,1:2,v_T) = wblfit(data2p(:,v_b,v_T));
        params2p(v_b,4,v_T)             = T_A;
        params2p(v_b,5,v_T)             = b_A;
        params2p(v_b,7,v_T)             = n;
       
        data3p(:,v_b,v_T) = wblrnd(v_T,v_b, [n,1]) + t0;
        start       = [1 0 0];
        custompdf   = @(x,a,b,c) (x>c).*(b/(a-c)).*(((x-c)/(a-c)).^(b-1)).*exp(-((x-c)/(a-c)).^b);
        opt         = statset('MaxIter',1e5,'MaxFunEvals',1e5,'FunValCheck','off');
        params3p(v_b,1:3,v_T) = mle(data3p(:,v_b,v_T),'pdf',custompdf,'start',start,'Options',opt,'LowerBound',[0 0 0],'UpperBound',[Inf Inf Inf])
        params3p(v_b,4,v_T)            = T_A;
        params3p(v_b,5,v_T)            = b_A; 
        params3p(v_b,6,v_T)            = t0;
        params3p(v_b,7,v_T)            = n;

    end
    result3p((v_T-1)*length(b)+1:v_T*length(b), 1:size(params3p, 2)) = params3p(:,:,v_T);
    result2p((v_T-1)*length(b)+1:v_T*length(b), 1:size(params2p, 2)) = params2p(:,:,v_T);
end

[Star Strider]

I am not exactly certain how to create the table this time.

Try this:

[vT,vb] = meshgrid(1:length(T), 1:length(b));
T2p = array2table(result2p);
T3p = array2table(result3p);
T2 = [table(vT(:), vb(:), 'VariableNames',{'v_T','v_b'}), T2p, T3p];

creating 3 tables and concatenating them, with:

First9Rows = T2(1:9,:)

producing:

First9Rows =

  9×16 table

    v_T    v_b    result2p1    result2p2    result2p3    result2p4    result2p5    result2p6    result2p7    result3p1    result3p2    result3p3    result3p4    result3p5    result3p6    result3p7
    ___    ___    _________    _________    _________    _________    _________    _________    _________    _________    _________    _________    _________    _________    _________    _________

     1      1       1.0287      1.0141          0            1            1            0           100        1.4767       0.96223      0.50666         1            1           0.5          100   
     1      2       1.0586      2.4242          0            1            2            0           100        1.5203        2.4382      0.43051         1            2           0.5          100   
     1      3       1.0107      3.6975          0            1            3            0           100        1.5031        2.6285      0.65671         1            3           0.5          100   
     1      4      0.96538      3.7811          0            1            4            0           100        1.5241        3.2105      0.73159         1            4           0.5          100   
     1      5       0.9754      5.0861          0            1            5            0           100        1.4811         5.481      0.37902         1            5           0.5          100   
     2      1       2.2783      1.0685          0            2            1            0           100        2.0837        0.8486       0.5487         2            1           0.5          100   
     2      2       2.0727      2.0381          0            2            2            0           100        2.6639        2.2925      0.54981         2            2           0.5          100   
     2      3       2.0826      3.1779          0            2            3            0           100         2.418        3.2345      0.38082         2            3           0.5          100   
     2      4       1.9133       3.394          0            2            4            0           100        2.3779         4.874      0.32657         2            4           0.5          100   

I am using those matrices because ‘param2p’ and ‘param3p’ are both (5x7x5) matrices, and I cannot figure out how to create them as 2D matrices to create the table object, and be consistent with the ‘v_T’ and ‘v_b’ subscripts.

.

[Mustafa Vural]

It looks really good, I really appreciate it. Thank you, you helped me a lot.

[Star Strider]

As always, my pleasure!