write table without nans to txt and csv

   writetable(t,'D.csv') and writetable(t,'D.txt')

are giving files with nan values. since the files are huge (around 1 million rows and 50 columns), when i replace nan with empty in these files the application crashes. the only way to deal with it is to allow matlab to write the table without nans. is it possible to do so?

[KSSV]

How you have that much huge data? I don't think writing nan's would be a problem.

[Danielle Leblance]

stata is having problem with the nans. i need the file with empty cells in order to process it with stata.

Answer 1

[Peter Perkins]

Danielle, if you can write to a spreadsheet, and then save the spreadsheet as csv, you'll get what you're looking for. Whether or not you can do that depends on what p[latform and version you're using.

You can't just "delete" the NaNs from a numeric variable, you'd have to convert to a cell array and replace cells containing NaN with empty, and you definately don't want to do that if you have a lot of data. You could however follow the lead of data analysts of days past, before NaN was a standard, and replace NaNs with something like -99 (or some other "impossible" value), save, load into Stata, and then deal with the -99's.

Hope this helps.

Answer 2

[Md Saifuddin]

I tried many ways to do that. The conversion table2cell and reconvert cell2table is very time consuming. The fastest way to deal with this, as I found is, to save the csv with 'NaN's in it. And then openning the file as text string and replace all the 'NaN's to your desired text. In my case I removed all my NaNs.

if true
  % code for 2016a
  writetable(Data,char(filename),'Delimiter',',');
  fid = fopen(char(filename),'rt') ;
  X = fread(fid) ;
  fclose(fid) ;
  X = char(X.') ;
  % replace string S1 with string S2
  Y = strrep(X, 'NaN', '') ;
  fid2 = fopen(char(filename),'wt') ;
  fwrite(fid2,Y) ;
  fclose (fid2) ;
end

Thanks to Jose for the replacing code portion https://www.mathworks.com/matlabcentral/answers/67052-how-to-modify-a-text-file-replacing-an-existing-string-with-an-user-defined-string#answer_118064

[Mustafa Fatih Çemen]

Hi when you replace NaN's with empty string, in text file two or more delimiters would be combind. For example "2,NaN,NaN,NaN,5" results in "2,,,,5" string, or "2,5,NaN" results in "2,5," How can we handle these situations such that they both result in 2,5 and 2,5 respectively.

Thanks.

Answer 3

[KSSV]

You may remove nan's like below:

k = rand(100,1) ; % an array
k(randsample(1:100,10)) = NaN ; % introduce some nan's
% remove nan's
k(isnan(k))=[];

[Danielle Leblance]

I received an error: Undefined function 'isnan' for input arguments of type 'table'.

[KSSV]

For table you cannot use isnan..can you copy a part of your table?