CSV file import with timestamp

I have csv data in the following format

value, timestamp, id
1.48,"2011-02-07 11:27:18+11",1

which I would like to import into Matlab so that I can look at the differences in successive time stamps. So I would like to convert the time of minutes to seconds and plot the difference. What is the best way to import this data?

Answer 1

[Marlies]

To import formatted data, textscan is usually a good option.

Textscan requires you to define a format. Assuming you want to separate all the elements in the file (value, every component from the timestamp separately, id-number), your format could look like this:

fmt = %f "%u-%u-%u %u:%u:%u+%u" %u

In this format; the '%f' and '%u' represent the datatypes that MATLAB is supposed to use for the data it reads (floating point, integer). All the other signs ("-:+) are 'literals'. They tell MATLAB that it will literally find a colon (that should be ignored).

In addition, you will need to define in the textscan-command that the delimiter is a comma

textscan(fid,fmt,'delimiter',',')

Hope this helps

Marlies

Answer 2

[Kye Taylor]

The assignment

fmt = %f "%u-%u-%u %u:%u:%u+%u" %u

should be

fmt = '%f "%u-%u-%u %u:%u:%u+%u" %u'

(notice the single quotes). To make the unpacking of output from textscan slightly easier, you could define fmt as below. Note the fopen and fclose statements before and after invoking textscan.

fmt = '%f%q%f';
fid = fopen('yourFileName.csv');
nowInMatlab = textscan(fid,fmt,'delimiter',',');
fclose(fid);