Erase all the rows that contain NaN

Dear all

I have

[N,T,R]=xlsread(name)

I want to erase all the rows in R that contain only NaN

thanks

[EDITED, copied from comments]:

So to input cell matrix is

 'TOTPAIN'            [NaN]                   [6.9150]
              [1x40 char]    [NaN]                  [6.90]
    [                NaN]    [NaN]                             [NaN]
    [                NaN]    [NaN]                       [NaN]
    'MAS'                   [NaN]                           [NaN]
    'MAR PACK G'        [NaN]                            [6.7]
    'MA 3 PACK 58G'        [NaN]              [6.7]

Answer 1

[Miro]

hi,

try this

[n,~]=size(R)
R_new = [];
for i = 1 : n
  R_tmp = R(i,:);
  if length(find(isnan(R_tmp)) ~= length(R_tmp) 
    R_new = [R_new;R_tmp]
  end
end

[Sabbas]

Hi Miro. I want to delete only the ROWS that contain only NaN.

By applying your command I obtain the following error message

Undefined function or method

'isnan' for input arguments of type 'cell'.

thanks

[Sabbas]

I am looking for something more compact Miro. Thanks anyway

Also, by applying your code I have

Error: Unbalanced or unexpected

parenthesis or bracket.

[Sabbas]

Just to remind you that R is a celll matrix

thanks

[Sabbas]

could anyone help?

Thanks

[Jan]

@Sabbas: Instead of reminding, it would be helpful if you explain such important details at first. Let me remind you, that the function xrsread is not a common function with known output. Do you mean xlsread?

[Sabbas]

Hi Simon. Yes, xlsread. I edited my question to correct this mistake So R is a cell matric that contains 'NaN', and other string variables like 'Hello' etc

[Miro]

sorry for the missunderstanding. the following should work Your example looks like you want to delete the Coloumn, not the Row... For deleting the column try the following:

      function Rout = deleteNAN(Rin)
      [~,n]=size(R)
      R_new = {};
      for i = 1 : n
          R_tmp = cell2mat(R(:,i));
          if length(find(isnan(R_tmp))) ~= length(R_tmp) 
            R_new = [R_new R_tmp]
          end
      end
      Rout = R_new

Store this in a .m file in your current folder and run it.

Answer 2

[Jan]

Perhaps something like this:

index = cellfun(@isequalwithequalnans, R, {NaN});
R(all(index, 2)) = [];

But due to a vague defined input, this contains too much guessing to be reliable. Sorry.

[Sabbas]

thanks simon. I applied you code and I obtain the following error message

Error using ==> cellfun All of the input arguments must be of the same size and shape. Previous inputs had size 376 in dimension 1. Input #3 has size 1.

So to input cell matrix is

 'TOTPAIN'            [NaN]                   [6.9150]
              [1x40 char]    [NaN]                  [6.90]
    [                NaN]    [NaN]                             [NaN]
    [                NaN]    [NaN]                       [NaN]
    'MAS'                   [NaN]                           [NaN]
    'MAR PACK G'        [NaN]                            [6.7]
    'MA 3 PACK 58G'        [NaN]              [6.7]

thanks

Answer 3

[Andrei Bobrov]

Rout = R(~all(cellfun(@(x)all(isnan(x)),R),2),:)

[Sabbas]

thanks Andrei. It does not work for the whole matrix. I get no warning but the rows with only NaN are not deleted

[Andrei Bobrov]

R = {...
    'TOTPAIN'          [NaN]    [6.915]
    'hfffgfjfjgfj'     [NaN]    [  6.9]
    [          NaN]    [NaN]    [  NaN]
    [          NaN]    [NaN]    [  NaN]
    'MAS'              [NaN]    [  NaN]
    'MAR PACK G'       [NaN]    [  6.7]
    'MA 3 PACK 58G'    [NaN]    [  6.7]};
>> Rout = R(~all(cellfun(@(x)all(isnan(x)),R),2),:)
Rout = 
    'TOTPAIN'          [NaN]    [6.915]
    'hfffgfjfjgfj'     [NaN]    [  6.9]
    'MAS'              [NaN]    [  NaN]
    'MAR PACK G'       [NaN]    [  6.7]
    'MA 3 PACK 58G'    [NaN]    [  6.7]
>>

[Jan]

@Sabbas: Do you get a "warning" or an "error"? In both cases posting the message would be a good idea.

[Sabbas]

Well, I get neither error nor warning. The rows full of NaNs still remain in the matrix. If I run the above codes you kindly provide for a matrix of 3 columns it seems to work