# remove Nan from a row vector using if statement only

4 views (last 30 days)
swetha S on 6 Jul 2019
Commented: madhan ravi on 6 Jul 2019
I have a row vector. Ch (1,1155). I want to remove columns that have NaN.
for k=1:122
p=1;
for m=1:1155
if( (ch(1,m) ~= NaN )
Datan(1,p)=ch(1,m);
p=p+1
end
end
end
But the above codes doesn't remove NAN

#### 1 Comment

Rik on 6 Jul 2019
As a side note: your outer loop does not affect the rest of your code, so it is useless.

madhan ravi on 6 Jul 2019
doc isnan

Guillaume on 6 Jul 2019
It's more than suffice, it's necessary!
madhan ravi on 6 Jul 2019
True , should have used "should" instead of "would" , thank you for the pointer :)

Guillaume on 6 Jul 2019
But the above codes doesn't remove NAN
One important property of NaN is that NaN is never equal to anything, including NaN. Hence NaN == NaN is always false, and NaN ~= NaN is always true.So, yes your code can't work.
As others have said, you have to use isNaN to check if a number is NaN. As usual in matlab, isNaN works on matrices too, so there's no point in using a loop:
datan = ch(~isnan(ch)); %as long as ch is a vector
It's unclear what the purpose of the k loop is in a code.
Finally, if you are hell-bent on using a loop, you should never hard code the end of a loop when you're looping over the elements of a matrix/vector. Always ask matlab how many elements there are. That way if the number of elements in the matrix/vector changes, you don't have to edit your code:
for m = 1:numel(ch) %iterate over the elements of ch

Stephan on 6 Jul 2019
Edited: Stephan on 6 Jul 2019
Datan=ch(~isnan(ch));