MATLAB: How To Efficiently Remove NaN Elements from Matrix

I am looking for a way to remove the NaN numbers from a matrix in MATLAB efficiently (i.e. without using a for loop)

I will provide a quick example to illustrate what I am trying to achieve:

Say I have a matrix M:

          3.00          1.00
          1.00          3.00
           NaN           NaN
          3.00          3.00
          1.00          1.00
           NaN           NaN
           NaN           NaN
           NaN           NaN
           NaN           NaN
           NaN           NaN

I would like to find a way to change this to

          3.00          1.00
          1.00          3.00
          3.00          3.00
          1.00          1.00

I am currently trying to do this via M(isfinite(M)) but that ends 开发者_高级运维up returning a vector instead of the matrix. Is there a trick to have it return a matrix instead?

If you have either no NaNs or all NaNs on each row, you can do the removal using:

M(isfinite(M(:, 1)), :)

The best way is

M(any(isnan(M),2),:)=[]

which will remove any row that contains at least one NaN.

Actually I would like to recommend a slightly different (and more general) approach.

So, in case that you want to ignore (i.e. delete) all the rows where at least one column includes NaN, then just:

M= M(0== sum(isnan(M), 2), :)

try my snip function. I wanted to address typical questions like this in one simple function:

B = snip(A,nan)

you can find the function file at

It also works with all other 'x', '0' or whatever elements and takes care of more similar problems.

The following function removes NAN from the data for specified dimensions:

function data_out = remove_nan (data_in, remove_dim)
%remove row or col from the data_in if there is a NaN element

% e.g., data_in =[1 2 3 4 NaN; 1 2 3 4 5; 1 2 3 NaN NaN]
% from this data remove col 4 and 5 such that data_out=[ 1 2 3; 1 2 3; 1 2
% 3]

if nargin==1

    col_loc=any(isnan(data_in),1);
    data_in(:,col_loc)=[];
    data_out=data_in;

elseif nargin==2

    if remove_dim=='col'
        %find the cols with nan and remove the colums
        col_loc=any(isnan(data_in),1);
        data_in(:,col_loc)=[];
        data_out=data_in;
    elseif  remove_dim='row'
        %find the rows with nan and remove the rows
        row_loc=any(isnan(data_in),2);
        data_in(row_loc,:)=[];
        data_out=data_in;
    end
else
    error( 'incorrect no of arguments')

end