Finding columns that contain both integers and NaN

3 views (last 30 days)
I have a matrix that contain Numerical values (float) and Categorical values (Int)
A = 1 NaN 2.2 3.2 4
NaN 7.9 5.1 NaN 5
3 5.5 NaN 4.1 NaN
and I'd like to split it into matrices;
one contains the Int values and the other contains the float values as
it = 1 4
NaN 5
3 NaN
flt = NaN 2.2 3.2
7.9 5.1 NaN
5.5 NaN 4.1
i use the following code
int_cols = all( round(A)==A, 1 ); %// logical indexing into integer columns
it = A(:,int_cols);
flt = A(:,~int_cols);
However, NaN isn't equal NaN so it appears in the flt matrix and it = [].
Are there any function that can help?

Answers (2)

David Young
David Young on 4 Feb 2016
Yes, isnan(x) does what x == NaN would do if NaNs were equal to each other.

the cyclist
the cyclist on 4 Feb 2016
int_cols = all( round(A)==A | isnan(A), 1 ); %// logical indexing into integer columns
it = A(:,int_cols)
flt = A(:,~int_cols)
  1 Comment
the cyclist
the cyclist on 4 Feb 2016
If the entries in A are have been computed in prior operations, you might also want to guard against floating point inaccuracies, such as the value "4" actually being 4.0000000000001. You could check for equality to within a specific tolerance, for example.

Sign in to comment.


Find more on Numeric Types in Help Center and File Exchange

Community Treasure Hunt

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

Start Hunting!

Translated by