0

I'm working on some MATLAB code that parses a .csv file into a table. The CVS file has seven columns separated by commas and also contains thousands of rows.

Right now the parsing code looks like this:

fid = fopen('data.csv', 'r');
parsed = textscan(fid, '%s %f %f %f %f %f %f', 'Delimiter',',', 'HeaderLines', 1);

But disp(parsed); returns an object that contains an array that has only one column. This data looks something like this:

[209,1] = 3/8/2015 16:12:00
[210,1] = 8.09
[211,1] = 952
[212,1] = 603
[213,1] = 100.8
[214,1] = 20.8
[215,1] = 11.3
[216,1] = 10.66
[217,1] = 3/8/2015 16:47:00
[218,1] = 8.1
[219,1] = 950
[220,1] = 604
[221,1] = 100
[222,1] = 15.8
[223,1] = 11.18
[224,1] = 10.71
[225,1] = 3/8/2015 17:22:00
[226,1] = 8.07
[227,1] = 981

How do I break this data up into a table. I want to be able to interface by data like: parsed[1][3]. I think I'm just missing an obvious parameter to pass into the textscan() function, but I can't find any documentation on this anywhere.

Any help you could provide would be greatly appreciated!

Update:

Here's a small sample of the CSV file I am working with:

dt,temp,ldo,turbidity,ldo.per,orp,conductivity,ph
3/8/2015 00:02:00,7.99,11.52,3.8,96.3,612,1038,8.01
3/8/2015 00:07:00,7.98,11.52,3.5,96.3,612,1038,8.01
3/8/2015 00:12:00,7.96,11.52,3.4,96.3,612,1038,8.01
3/8/2015 00:17:00,7.97,11.54,3.7,96.5,612,1038,8.01

I'd like the output to look something like:

A =
   3/8/2015 00:02:00   7.99   11.52   3.8   96.3   612   1038   8.01
   3/8/2015 00:07:00   7.98   11.52   3.5   96.3   612   1038   8.01
   3/8/2015 00:12:00   7.96   11.52   3.4   96.3   612   1038   8.01
   3/8/2015 00:17:00   7.97   11.54   3.7   96.5   612   1038   8.01
James Taylor
  • 6,158
  • 8
  • 48
  • 74

1 Answers1

2

You can use importdata:

d = importdata('filename.csv');

This gives a struct d with fields data (numeric) and textdata (cell array of strings):

>> d.data
ans =
  1.0e+003 *
    0.0080    0.0115    0.0038    0.0963    0.6120    1.0380    0.0080
    0.0080    0.0115    0.0035    0.0963    0.6120    1.0380    0.0080
    0.0080    0.0115    0.0034    0.0963    0.6120    1.0380    0.0080
    0.0080    0.0115    0.0037    0.0965    0.6120    1.0380    0.0080

>> d.textdata
ans = 
    'dt'                   'temp'    'ldo'    'turbidity'    'ldo.per'    'orp'    'conductivity'    'ph'
    '3/8/2015 00:02:00'    ''        ''       ''             ''           ''       ''                ''  
    '3/8/2015 00:07:00'    ''        ''       ''             ''           ''       ''                ''  
    '3/8/2015 00:12:00'    ''        ''       ''             ''           ''       ''                ''  
    '3/8/2015 00:17:00'    ''        ''       ''             ''           ''       ''                ''

To combine them you need a cell array:

result = [d.textdata(2:end,1) num2cell(d.data)];

which gives

result = 
    '3/8/2015 00:02:00'    [7.9900]    [11.5200]    [3.8000]    [96.3000]    [612]    [1038]    [8.0100]
    '3/8/2015 00:07:00'    [7.9800]    [11.5200]    [3.5000]    [96.3000]    [612]    [1038]    [8.0100]
    '3/8/2015 00:12:00'    [7.9600]    [11.5200]    [3.4000]    [96.3000]    [612]    [1038]    [8.0100]
    '3/8/2015 00:17:00'    [7.9700]    [11.5400]    [3.7000]    [96.5000]    [612]    [1038]    [8.0100]
Luis Mendo
  • 110,752
  • 13
  • 76
  • 147
  • Thanks for the answer! I had to change your code slightly to involve the delimiter. So I have: `d = importdata('data.csv', ',', 1);` but the `1` doesn't seem to be truncating the header. When I run it, I get `error: horizontal dimensions mismatch (1836x1 vs 1736x7)` which makes sense since the first line of `textdata` is the header. Any way around this? – James Taylor Apr 07 '15 at 15:34
  • I don't see how `d.textdata` can have 1836 rows and `d.data` only 1736. Perhaps some of your numeric data is not being read as numbers – Luis Mendo Apr 07 '15 at 16:40
  • I miss typed that. It was an off by one error, but it turns out that one line `3/14/2015 NA,NA,NA,NA,NA,NA,NA,NA` was throwing it off, so you were right. As a side question, is there any way to account for this beforehand? – James Taylor Apr 07 '15 at 16:53
  • @JamesTaylor I don't know... In your case, perhaps you could detect which rows of `d.textdata` don't have a `d.data` counterpart by seeing which rows of `d.textdata` contain characters other than numbers or `/`, and then remove those rows – Luis Mendo Apr 07 '15 at 16:59