CSV import with variable column headers

Hey there, I'm new to MATLAB,and I've searched the forums and MATLAB help and can't seem to find an straightforward way to create a generic CSV import script. I'm looking to take a generated CSV file after a test run where the number of parameters can vary from test to test and import it to MATLAB for post processing.

The format of the CSV file is consistent with a format as such:

Test Name
Test Date
Start time
Parameter ID1, PID2, PID3, ... PIDn
EU1, EU2, EU3, ... EUn
Data 1, data 2, data 3, ... Data n

I know that I want to take row say 6 and make that my variable names, and row 8 is the start of the data, but column width (e.g. channels) could change depending on test conditions.

Cheers

Answer 1

[Felix]

It should be noted that importdata is an extremely convenient alternative function for this and auto-detects the column headers.

Answer 2

[Geoff Hayes]

Rather than creating n variables for each column of data, why not just create a mxn matrix where n is the number of columns in your file. Reading the CSV data becomes easy

data = csvread('test.txt',8);

where 8 is the row that you want to start reading the data from. The first seven (header) rows are ignored and you can now do any post processing on data. (This assumes the format of the file is strictly the header of seven lines followed by m rows of numeric data.)

[Lance]

Thanks! Is there a way to just extract the Parameter IDs using something like textscan then parse those strings with the associated data matrix?

[Geoff Hayes]

Note that in the above command, csvread, the row argument is zero-based. So if you want the 8th row, you must supply 7 instead of the 8 that I mention.

If you just want the Parmeter IDs that are in say row 6, then you could do the following

 fid      = fopen('test.txt');
 paramIds = textscan(fid,'%s',1,'HeaderLines',5);
 fclose(fid);

paramIds will be a single-element cell array with the contents of the sixth line from your file (so comma-separated line of parameter ids). In the inputs to textscan, we indicate that we just want one row with 1, and that there are 5 header rows to skip.

For more info on these commands, type doc textscan and doc csvread.