Read and write csv files without changing properties

I've succesfully solved a big problem thanks to you guys and now I'm facing a slightly smaller but important one.

(Following code is the original one posted by another mvp here. Kudos)

data = int64(csvread('Sample_02.csv')); 
time = data(:,2);

% remove empty rows
data = data(time>0, :);
time = time(time>0);

time = mod(time, 1000000); % yyyymmdd are not important so discard them
current_time = 00000;   % time in hhmmss format

while size(data, 1) > 0
    next_time = current_time + 30000; % 30000 represent 3 hours
    index = time < (current_time + 30000);
    partial_data = data(index, :);
       
     dlmwrite(['data-' num2str(current_time/10000) '-' num2str(next_time/10000) '.csv'], ...
         partial_data, 'precision', '%i');
    
    data(index,:) = [];
    time(index,:) = [];
    current_time = current_time + 30000;
end

The code reads the sample CSV file and split it in to seperate csv files based on time-stamps.

My problem is: The original csv file (Sample_02.csv) has 7 columns and the above code only deals with the second one. Output files also have 7 columns, obviously, but the values are not shown/stored as decimals. For example, the original CSV file stores 123.456789 in its 6th column but the generated output files have only 123 in the respective 6th column.

How can I store the same data as decimals in the output files?

TIA!

[Walter Roberson]

Don't use 'precision', '%i' -- the %i means to only write out integers.

Preserving the same number of decimal places as the original requires processing the lines as text rather than numeric.

[Jake]

Thank you!

But I did try removing 'precision', '%i' from the code but it only affects the first two columns of the output csv file. Also, output comes as a power (ex: 2.02002E+13 ) and not as the data stored in the original file. (Sample_02.csv)

[Walter Roberson]

Preserving the same number of decimal places as the original requires processing the lines as text rather than numeric.

[Jake]

So I should go with something other than csvread?

I'm sorry about the noob questions, I'm still quite new.

[Mohammad Sami]

I tested your sample file with readtable and writetable. It seems to work as you desire. I am using R2019b.

[Jake]

You mean the output data were in the format with decimals? I'm using R2017a and it didn't work for some reason :(

Answer 1

[Ameer Hamza]

Hi James, I posted the answer to your other question. You can just replace the line

data = int64(csvread('Sample_02.csv')); 

to

data = csvread('Sample_02.csv');

and it will retain the decimal part of the numbers.

I initially used int64 to preserve precision in the 2nd column, but since it has 14 digits, so it can be handled without int64 too.

[Jake]

Yes! I did this and it works.

Answer 2

[Walter Roberson]

fileread to read the file into a string.

textscan the string to extract the datetime as numeric.

Now regexp the string with '\r?\n', 'split' to create a cell array in which each entry is a line.

Use the information in the numeric date to figure out where to split. The indices can be used in the cell array of split lines. strjoin() those cells with '\n' and fwrite() the result.

[Jake]

Thank you, Walter.

I'm still new so it's a bit complicated for me after the second line of suggestion. But I will try to work around.

[Walter Roberson]

S = fileread(filename);
fmt = '%*f,%ld,%*s';
C2 = cell2mat(textscan(S, fmt));
lines = regexp(S, '\r?\n', 'split');
time = double(mod(C2, 1e6));

 ...
  mask = start_time <= time & time < end_time;
  subset = strjoin(lines(mask), '\n') ;
  fwrite(fid, subset);

[Jake]

Thank you!