Replace String with a NaN in table

Hey guys!

I am challenged with a table, consisting various numbers and a few "'Bad'" cells. I guess they are strings. My goal now is to replace these "Bads" with NaNs.

How do I achieve that the easiest way?

Best regards, Spacey

[per isakson]

"table" &nbsp is that

• table, Create table from workspace variables
• a Cell Arrays or
• a file

Answer 1

[Peter Perkins]

It sounds like you imported a file where some of the columns were mostly numbers, but some strings mixed in. Best to avoid getting into the corner that you're in. You might have been able to explicitly say that the column is numeric. You might have been able to use the TreatAsEmpty parameter to readtable (if that's what you used).

Otherwise, if you have variables in the table that are cell arrays but that you want to be numeric, you might need to do one of a couple different things. In general, you need to put numeric things in all the cells and then concatenate the contents to get a numeric vector, but first you'll need to describe exactly what you have.

Answer 2

[Konstantinos Sofos]

Hi,

" I guess they are strings" ...you cannot open your table or see in the workspace what type are they?

Do you mean something like the following:

>> A = {'xx',5,6,'''',78,'?..'}

A = 

      'xx'    [5]    [6]    '''    [78]    '?..'

>> idx = cellfun(@(x) ischar(x),A)

idx =

       1     0     0     1     0     1

>> A(idx)={NaN}

A = 

      [NaN]    [5]    [6]    [NaN]    [78]    [NaN]

>> B = cell2mat(A)

B =

     NaN     5     6   NaN    78   NaN

>> whos
  Name      Size            Bytes  Class     Attributes

    A         1x6               408  cell                
    B         1x6                48  double

Answer 3

[Jos (10584)]

Take a look at CELL2FLOAT

http://www.mathworks.com/matlabcentral/fileexchange/19730-cell2float

Answer 4

[Spacey]

Unfortunately I couldn't get it working yet. The huge problem is the table. It says that the "'Bad'" values are cells, but many functions don't even work with cells respectively different formats within the table. Do guys have any other ideas or questions that would specify this problem?