Extracting rows from csv file which contain only certain text

Hello, I have a csv file containing all the information I need. However, there are some rows I will not use in my data analysis. So I need to take out only certain rows I need, in order to have proper file to work with. The question is: If i want to extract all the rows which contain specific text in one of their columns, for instance 'Kitchen', and save only them to a new csv file, how do i do that?

Answer 1

[dpb]

doc strfind

NB: there's no specific reason to make a new file unless you have other reasons besides doing the analysis; just read the data and select the desired subset in memory and do the analysis...certainly no need to

Open/Read/Close/Select/CreateOpenNew/Write/Close/Open/ReadSelected/Close/Analyze

when can simply

Open/Read/Close/Select/Analyze

[N G]

Thanks for the answer, I'm quite new to MatLab. I do need another file cause I will later do periodic data out of it, so i dont want to get lost in unnecessary data as well, I want to simplify this to myself. With strfind I obtain the vector showing the pattern of the chosen string to be found. But to be sincere I dont know how to later use this vector to obtain a new file consisting only of the rows which include my wanted string. Is there possibility that you write me the example of script for this? Kind regards :)

[dpb]

Given data array x containing the string from which you wish to find a particular match in one column it's a simple as

x=x(strfind(x(:,3),'kitchen'),:);  % say it's column 3, fixup as needed

will leave you with the array containing only the rows in which there was the match. NB: strfind is case-sensitive so you either need to match the precise string or do something like

x=x(upper(strfind(x(:,3)),'KITCHEN'),:);

To write depends on the format you want and the content; the easy-to-use csvwrite will only operate on numeric data so you will probably have to use fprintf unless TMW has introduced some new, more generic i/o routines since R2012b which is the latest I have here.

Hope that helps...

[Image Analyst]

Well, if you have a table variable, with some columns being strings and other columns being numbers, you can use writetable().

fprintf() is the most flexible though, but it take a few more steps than writetable.