MATLAB Answers

How do I remove rows with NaNs from my raw data?

12 views (last 30 days)
Tessa
Tessa on 18 Nov 2014
Commented: Guillaume on 25 Nov 2014
I have opened my excel data with xlsread and would like to remove all rows containing a NaN. Is there a possibility to do this at once? my data looks like this:
information blabla NaN NaN information blabla NaN NaN
NaN NaN NaN
Mydata Mydata Mydata My Data Mydata Mydata Mydata My Data Mydata Mydata NaN My Data Mydata Mydata Mydata My Data Mydata Mydata Mydata NaN Mydata Mydata Mydata My Data
It´s a cell array in case that matters :)
  1 Comment
Guillaume
Guillaume on 18 Nov 2014
It does matter greatly that it's a cell array. Other than the NaNs, what's in the cell array? strings?, matrices? or just scalars?

Sign in to comment.

Answers (2)

Giorgos Papakonstantinou
Giorgos Papakonstantinou on 18 Nov 2014
I assume that your data are of nx1 or 1xn size. Then to delete the use this:
data(isnan(data)) = [];
  4 Comments
Tessa
Tessa on 25 Nov 2014
Thanks for the response. It is a mixture of numbers, dates and text. So both are not working... Is there a way to change the cell array in something numeric eventhough it is not all numbers?

Sign in to comment.


Guillaume
Guillaume on 25 Nov 2014
Tessa, try this:
c={'sometext', NaN, 12.5
'othertxt', 45, 12
NaN, 'blah', 7
48, 78, 'aaa'} %example
nanrows = any(cellfun(@(e) isnumeric(e) && isnan(e), c), 2);
c(nanrows, :) = []
It will work as long as you don't have a matrix of numbers in a cell. If you do, replace the
isnan(e)
by
any(isnan(e(:)))
  2 Comments
Guillaume
Guillaume on 25 Nov 2014
If you get an empty cell, that means each row must contain at least one NaN.
I find it a bit odd to discard a whole row if there's just one NaN in it, but that's what you requested: "remove all rows containing a NaN".
The @(e) isnumeric(e) && isnan(e) is an anonymous function where e is just a placeholder name for any cell of the cell array. This is equivalent to:
function out = anonymousfunction(e)
out = isnumeric(e) && isnan(e);
end
Obviously, you could use any name you want instead of e.

Sign in to comment.

Tags

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!