Deleting runs of zeros

Hello! I have a dataset that occasionally has runs of zeros. I would like to delete those runs from my data. How do I state that for any run of 3 or more zeros in my data, I would like Matlab to completely delete those observations from my dataset.

So for example, I have a dataset that looks like

[1 5;2 7;3 0;4 0;5 0;6 3]

I would like the that run of zeros deleted so that my new dataset looks like...

[1 5;2 7;6 3]

Any help is greatly appreciated!!

Answer 1

[Guillaume]

one possible way, use strfind (which works with numbers despite the name). Using a more complex example (to show that it works regardless of the numbers of 0 in the run):

data = [1 5; 2 0; 3 0; 4 2; 5 3; 6 0; 7 0; 8 0; 9 0; 10 1; 11 0; 12 0; 13 0]
startseq = strfind(data(:, 2)', [0 0 0]);
toremove = unique([startseq, startseq+1, startseq+2]);
data(toremove, :) = []

[Lauren]

Thank you very much!

Answer 2

[Doug]

a = [1 5;2 7;3 0;4 0;5 0;6 3];

i = a(:,2)~=0;

a = a(i,:);

[Lauren]

This code removes all zeros from my dataset. I only want to remove runs of 3 or more consecutive zeros. Can you modify this code to reflect that?

Answer 3

[Joseph Cheng]

you would find the indexes that equal zero then if the number is greater or equal to three then delete them from original like in the example below;

 x = [1 5;2 7;3 0;4 0;5 0;6 3]

[r c v]=find(x==0);

if numel(r)>=3
    x(r,:)=[];
end

[Stephen23]

@ Joseph Cheng: do not put empty lines in your code, and then this page will format the entire code in one box.

Also while this certainly detects zeros, it does not detect runs of zeros, which by definition must be consecutive. This code does not resolve this issue.

[Joseph Cheng]

?when did the page layout change? oh and yes i didn't see the comment on runs of zeros. However quick modification of taking the diff of r (diff(r)) and looking for values of 1, one can determine runs of zeros.