Delete all rows in cell array based on value

I have a cell array A that contains multiple tables in the first column and add a vector (B) in the second column that shall indicate if a row is used or can be deleted.

A = {table(rand(3,2)); table(rand(3,2)); table(rand(3,2))};

B = {1; 0; 1};

C = horzcat(A,B);

I now have issues to delete all rows in cell array C that are indicated with a 0.

I have tried some proposed solutions for this kind of problem (deleting rows based on value) but could not find a working one for the specific problem.

Has someone an idea how to handle this issue?

Best regards!

Answer 1

[J. Alex Lee]

Do you need the cell array C? Must B be a cell array? If not...

A = {table(rand(3,2)); table(rand(3,2)); table(rand(3,2))};

% make B a logical array
% B = {1; 0; 1};
B = ([1;0;1]);

% don't need C
% C = horzcat(A,B);
A(~B) = [];

If for whatever reason you need an inline solution once you have the cell array C

A = {table(rand(3,2)); table(rand(3,2)); table(rand(3,2))};
B = {1; 0; 1};
C = horzcat(A,B);

C(~C[{:,2}],:)=[]

[Veronica Taurino]

Your solution is way better than mine! Thank you, I am going to save it. However it doesn't run on my Matlab, I changed it a bit:

A = {table(rand(3,2)); table(rand(3,2)); table(rand(3,2))};
B = [1; 0; 1];
C=A;

C(~B(:))=[];

[Image Analyst]

Veronica, Alex's answer runs just fine - I actually tried it. It's how I would have done it if I wanted to change A in place.

If you wanted a separate output variable C, then you can do what you did but you don't need (:), so do it like this (which I also tested):

C = A; % Initialize.
C(~B) = [] % Remove elements where B is zero.

[Veronica Taurino]

Could do help me to understand why it does not run on mine?

A = {table(rand(3,2)); table(rand(3,2)); table(rand(3,2))};
B = {1; 0; 1};
C = horzcat(A,B);
C(~C[{:,2}],:)=[]

C(~C[ {:,2} ],:)=[]

↑

Error: Unbalanced or unexpected parenthesis or bracket.

I get this. I have Matlab R2017a. Thank you

[Jan Kielmann]

Exactly what I was looking for! Thanks to both of you. You are correct, I don't need to write the logical array into the data cell array. I wanted to be sure that the issue was clear.

[Image Analyst]

Veronica, C is a cell array and thus does not use square brackets for indexing. It only uses braces or parentheses. I think once you read the FAQ it should become clear to you. It will give you a good intuitive feel for what a cell array is and how to use it.

And horzcat() is not needed at all because there was never any need to stitch B onto the end of A.

[Veronica Taurino]

I know this. Maybe there is a misunderstanding here. You said Alex's answer runs just fine, but I get that error using his solution (the second part of his answer). For this reason I was asking how it could run. I am not a native english speaker, there is a good chance my question is not clearly written, sorry about that. Thank you

[J. Alex Lee]

It's my fault, I had a typo in the 2nd answer, sorry about that!

C(~[C{:,2}],:)=[]

This is just "undoing" the redundant cell-ification and horzcat of B onto A; just another way of saying what Image Analyst said above.

In more detail:

The FAQ above doesn't appear to succinctly cover what I was trying to do here. Multiple indexing of a cell array using curly braces returns multiple values, where as multiple indexing of a cell array using parantheses returns the subset of the cell array as another cell array

A = {table(rand(3,2)); table(rand(3,2)); table(rand(3,2))};
B = {1; 0; 1};
C = horzcat(A,B);

C(:,2)

returns

ans =

  3×1 cell array

    {[1]}
    {[0]}
    {[1]}
    

whereas

C{:,2}

returns

ans =

     1


ans =

     0


ans =

     1
     

You can "catch" the multiple results of the latter by wrapping in square brackets, as if defining an array manually

[C{:,2}]

returns

ans =

     1     0     1
     

[Veronica Taurino]

Thank you Alex, I was going crazy moving those brackets around ahah neat solution, thank you, I am going to save it for later. Have a nice day

[J. Alex Lee]

through browsing more on answers, I found this:

https://www.mathworks.com/help/matlab/matlab_prog/comma-separated-lists.html

Actual documentation about what I was doing with [C{:,2}]

Answer 2

[Veronica Taurino]

A = {table(rand(3,2)); table(rand(3,2)); table(rand(3,2))};
B = {1; 0; 1};
C = horzcat(A,B);

D={};
ii=0;
jj=0;
for ii=1: size(C,1)
    if C{ii,2}==1
       jj=jj+1;        
       D{jj}=C{ii,1}; 
       D=D';
    end
end

Something like that?