Thread Subject:

extract rows

I have the following script...

j = [time data]; %8760*4 matrix
m = maxtab(:,1); %Column of data
rows = find(ismember(time, m,'rows')); %find those rows where m value matches time value
j(rows,:)=[];

What I am trying to do is extract those lines that correspond to the m value from matrix j. The last line deletes the rows in J that correspond to m. However, what I need is the opposite, which is delete all other values and retain only those J rows that correspond to the m values.

Thanks for your help,
Jon

"Jonathan"
> j = [time data]; %8760*4 matrix
> m = maxtab(:,1); %Column of data
> rows = find(ismember(time, m,'rows')); %find those rows where m value matches time value
> j(rows,:)=[];
> What I am trying to do is extract those lines that correspond to the m value from matrix j. The last line deletes the rows in J that correspond to m. However, what I need is the opposite, which is delete all other values and retain only those J rows that correspond to the m values...

a nice example of the strength of
- logical indexing, which is what you want
and
- linear indexing, which is what FIND unnecessarily does for you in this case

one of the solutions

     d=[1,3,4,5,7,9];
     t=[1,7,2];
     ix=ismember(d,t)
% ix = 1 0 0 0 1 0
     dok=d(ix)
% dok = 1 7
     dnotok=d(~ix)
% dnotok = 3 4 5 9

us

"us " <us@neurol.unizh.ch> wrote in message <gq1b6t$bf3$1@fred.mathworks.com>...
> "Jonathan"
> > j = [time data]; %8760*4 matrix
> > m = maxtab(:,1); %Column of data
> > rows = find(ismember(time, m,'rows')); %find those rows where m value matches time value
> > j(rows,:)=[];
> > What I am trying to do is extract those lines that correspond to the m value from matrix j. The last line deletes the rows in J that correspond to m. However, what I need is the opposite, which is delete all other values and retain only those J rows that correspond to the m values...
>
> a nice example of the strength of
> - logical indexing, which is what you want
> and
> - linear indexing, which is what FIND unnecessarily does for you in this case
>
> one of the solutions
>
> d=[1,3,4,5,7,9];
> t=[1,7,2];
> ix=ismember(d,t)
> % ix = 1 0 0 0 1 0
> dok=d(ix)
> % dok = 1 7
> dnotok=d(~ix)
> % dnotok = 3 4 5 9
>
> us

Hmmm, while this approach seems better than using find, it still does not resolve my original problem.

If my original matrix 'd' in your example above is 8760*4, and 't' is 9*1, what I would like to be able to do is for the 9 't'-values, use these to identify the same row values in 'd' and place those specific 'd'-rows in a new matrix say 'rows', which would have a size of 9*4.

Thanks again for your help.

Jon

"Jonathan" <jkakiwi@yahoo.co.uk> wrote in message <gq89o5$d6n$1@fred.mathworks.com>...
> Hmmm, while this approach seems better than using find, it still does not resolve my original problem.
>
> If my original matrix 'd' in your example above is 8760*4, and 't' is 9*1, what I would like to be able to do is for the 9 't'-values, use these to identify the same row values in 'd' and place those specific 'd'-rows in a new matrix say 'rows', which would have a size of 9*4.
>
> Thanks again for your help.
>
> Jon

  Urs has shown you the basic technique, Jonathan. Just apply his method to your particular problem. Don't use the 'rows' version of 'ismember' since you are only doing matching with the first column of j.

 p = ismember(j(:,1),m);
 j = j(p,:);

Roger Stafford

"Roger Stafford" <ellieandrogerxyzzy@mindspring.com.invalid> wrote in message <gq8fc5$icq$1@fred.mathworks.com>...
> "Jonathan" <jkakiwi@yahoo.co.uk> wrote in message <gq89o5$d6n$1@fred.mathworks.com>...
> > Hmmm, while this approach seems better than using find, it still does not resolve my original problem.
> >
> > If my original matrix 'd' in your example above is 8760*4, and 't' is 9*1, what I would like to be able to do is for the 9 't'-values, use these to identify the same row values in 'd' and place those specific 'd'-rows in a new matrix say 'rows', which would have a size of 9*4.
> >
> > Thanks again for your help.
> >
> > Jon
>
> Urs has shown you the basic technique, Jonathan. Just apply his method to your particular problem. Don't use the 'rows' version of 'ismember' since you are only doing matching with the first column of j.
>
> p = ismember(j(:,1),m);
> j = j(p,:);
>
> Roger Stafford

Thanks Urs and Roger. This was extremely helpful.