convert char to double

Hi, I have a data file with strings. It looks like this with 4 columns;

data=19-Jan-2015 11:30:00.000    11.8950011     1.7625011     1.7481285 

19-Jan-2015 11:30:01.000    11.8965768     1.7640768     1.7496914 

19-Jan-2015 11:30:02.000    11.8952753     1.7627753     1.7484005

i need to separate these columns for my further analysis. i tried y=str2num(data) but it didn't work. Any advice please. Thank you Darshani

[Walter Roberson]

Is there a tab delimiter? Visually it appears to be 5 columns, as 19-Jan-2015 would be a different column then 11:30:01.000 if you are using space delimiter.

Do you need the date/time converted to numeric form?

Which MATLAB version do you have? In particular, R2014b or later, or an earlier version?

Answer 1

[Binu]

Hi Walter, it is 4 columns and the first column has no spaces in it. Yes I need to convert date + time into numeric format. Thank you Darshani

[Walter Roberson]

Please do not start a new Answer to reply. You can click on "Comment on this Answer" to reply to someone.

[Walter Roberson]

Consider your sample input line

19-Jan-2015 11:30:01.000    11.8965768     1.7640768     1.7496914

This line has 3 clear numeric values, 11.8965768, 1.7640768, and 1.7496914. That is 3 columns. But you also have

   19-Jan-2015 11:30:01.000

which contains a space in it. You say that none of your columns have spaces in them, so that implies that one column is 19-Jan-2015 and another column is 11:30:01.000. That's 2 columns. 2 columns plus the 3 clear numeric columns gives 5 columns, not 4.

If you have a day-month-year and a time and you have 4 columns with no spaces, then you could only fit two numeric columns , not 3.

We cannot tell you how to read in the data and convert it until we know exactly what the lines look like.

[Binu]

Hi Walter, Here I attch a part of my data file (same file in .mat and .zip) which would help you see how the data looks like. Thank you Darshani

Answer 2

[Rob Purser]

Assuming you're in R2014b or later:

Recreate your data

    data = ['19-Jan-2015 11:30:00.000    11.8950011     1.7625011     1.7481285 ' 13 '19-Jan-2015 11:30:01.000    11.8965768     1.7640768     1.7496914' 13 '19-Jan-2015 11:30:02.000    11.8952753     1.7627753     1.7484005']

Parse using textscan

%%Use textscan to parse
C = textscan(data, '%s %s %f %f %f');

% reorganize the cell array
timestamp = strcat(C{1},{' '},C{2});
timestamp = datetime(timestamp,'InputFormat','dd-MMM-yyyy HH:mm:ss.SSS');
parseddata = [C{3} C{4} C{5}];

timestamp
parseddata

Output:

timestamp = 

     19-Jan-2015 11:30:00
     19-Jan-2015 11:30:01
     19-Jan-2015 11:30:02

parseddata =

     11.8950    1.7625    1.7481
     11.8966    1.7641    1.7497
     11.8953    1.7628    1.7484

Answer 3

[Binu]

Thanks Walter and Rob.I have R2012b.Any more advice suit for this version? Thanks Darshani

[Walter Roberson]

Is the data 5 columns space-separated, or is it 4 columns tab-separated with the first column happening to have a space in it?

Do you need the date + time converted to numeric form?

Answer 4

[Walter Roberson]

lines_cell = regexp(regexprep(Test_data, '\s+$', ''),'\s+', 'split');
lines = vertcat(lines_cell);
numeric_vals = str2double(lines(:,3:end));
dates = datenum(strcat(lines(:,1), {' '}, lines(:,2)), 'dd-mmm-yyyy HH:MM:SS.fff');

Now dates(K) relates to numeric_vals(K,:) . You can break numeric_vals out into separate columns if that is useful.

[Binu]

Great! Thanks Walter