Warning: The assignment added rows to the table, but did not assign values to all of the table's existing

I have the following code that generates the Warning

nmonths = 100; % use 100 months worth of data to estimate the betas

for i = 1:nstocks

for j = 0:(nmonths-60) % 5 x 12 = 60 months of estimation window and

% obtain 41 observations of beta for each stock i

mdlbetas = fitlm(monthlyX((1+j):(60+j),:),monthlyY((1+j):(60+j),i));

Betas(i,j+1) = mdlbetas.Coefficients(2,1);

end

end

I get the warning: Warning: The assignment added rows to the table, but did not assign values to all of the table's existing

How do I avoid this warning and should I even worry about this if I get the output I desire?

Answer 1

[Voss]

The warning happens because the table Betas gets a new row added each time the j loop runs its first iteration (except when i is 1, since in that case there are no variables in the table Betas yet), but only the value of the first column/variable in the new row is set (the rest default to 0).

nmonths = 62;
nstocks = 2;
monthlyX = rand(nmonths,1);
monthlyY = rand(nmonths,nstocks);
Betas = table();
for i = 1:nstocks
    for j = 0:(nmonths-60) % 5 x 12 = 60 months of estimation window and 
        % obtain 41 observations of beta for each stock i
        mdlbetas = fitlm(monthlyX((1+j):(60+j),:),monthlyY((1+j):(60+j),i));
        fprintf('i = %d, j = %d:',i,j);
        Betas(i,j+1) = mdlbetas.Coefficients(2,1);
        disp('Betas = ');
        disp(Betas);
    end
end

i = 1, j = 0:

Betas = 

      Var1  
    ________

    0.065973

i = 1, j = 1:

Betas = 

      Var1        Var2  
    ________    ________

    0.065973    0.058035

i = 1, j = 2:

Betas = 

      Var1        Var2        Var3  
    ________    ________    ________

    0.065973    0.058035    0.040655

i = 2, j = 0:

Warning: The assignment added rows to the table, but did not assign values to all of the table's existing variables. Those variables are extended with rows containing default values.

Betas = 

      Var1         Var2        Var3  
    _________    ________    ________

     0.065973    0.058035    0.040655
    -0.083345           0           0

i = 2, j = 1:

Betas = 

      Var1         Var2        Var3  
    _________    ________    ________

     0.065973    0.058035    0.040655
    -0.083345    -0.10717           0

i = 2, j = 2:

Betas = 

      Var1         Var2        Var3  
    _________    ________    ________

     0.065973    0.058035    0.040655
    -0.083345    -0.10717    -0.13571

You can avoid this warning by initializing the table Betas to have the full number of rows and columns it will need.

However, maybe Betas doesn't need to be a table at all; maybe it can be a numeric matrix:

Betas = NaN(nstocks,nmonths-60+1);
for i = 1:nstocks
    for j = 0:(nmonths-60) % 5 x 12 = 60 months of estimation window and 
        % obtain 41 observations of beta for each stock i
        mdlbetas = fitlm(monthlyX((1+j):(60+j),:),monthlyY((1+j):(60+j),i));
%         Betas(i,j+1) = mdlbetas.Coefficients(2,1);
        Betas(i,j+1) = mdlbetas.Coefficients{2,1};
    end
end
disp(Betas);

    0.0660    0.0580    0.0407
   -0.0833   -0.1072   -0.1357