nan problem

any ways to remove NAN in the final results

[Walter Roberson]

Duplicate is at http://www.mathworks.com/matlabcentral/answers/16221-nan-value

Answer 1

[Jan]

A very vague question. I guess the answer is: yes.

x = [1, NaN, 3];
x(isnan(x)) = [];

[x]

when this is done then ahat will be the response sir?nan values wont be displayed..but this may affect the performance ah

[x]

what does 1,3 means

[Wayne King]

Jan is giving you an example by creating a vector,x, with one NaN. Then he is using isnan() to find the element of x that is a NaN and removing it.

[x]

FOR A 9 BY 9 MATRIX I AM GETTING NAN value..i am getting complex term as 9by9 matrix so i used normalization but for few patches i come across 0/0 so the NAN returns..what can i do for this sir

[Jan]

@x: An explicite example would be much easier to understand.

If your results contain some 0/0==NaN values, it is your turn to define what should happen in this case. It is impossible for us to guess, what a good replacement could be, because this depends on your problem.

Perhaps: x(isnan(x)) = 0, or x(isnan(x)) = -1000, or whatever.

If you plot the results NaN's are often treated very handy, because they are not drawn. See:

plot([1:10, NaN, 11:16]);

[Andrei Bobrov]

Hi Jan!

x = x(~isnan(x))