replace missing values in a matrix

Consider the following:

A = [1,1,1;
    2,3,1;
    nan,3,1;
    nan,3,nan];

how would I replace the missing values with the previous values? For example:

data =

       1     1     1
       2     3     1
       2     3     1
       2     3     1

I can find the nan by isnan(data) but I don't know how to replace the values as described.

Amended:

I've managed to do this but not in a very concise way:

A = [1,1,1;2,3,1;nan,3,1;nan,3,nan];
B = ~isnan(A);
I = arrayfun(@(x)find(B(:,x),1,'last'),1:size(A,2));

for i = 1:size(A,2);    
    C = A(:,i);
    D = isnan(C);
    C(D) = C(I(i));
    Final(:,i) = C;
end

Could anyone suggest on a way to tidy this up?

[Jan]

Is this matrix a possible input:

A = [1,   NaN; ...   % NaNs in first row
     NaN, 2; ...     % Non-NaNs after a NaN
     3,   NaN]

Answer 1

[Jan]

What about a simple loop:

• Go through the rows from 2 to end
• Use isnan() to find the indices of NaNs
• Replace values in the current row by the ones of the former row.

I'm not more specific, because this could be a homework also.

[EDITED]

for k = 2:size(A, 1)
  index       = isnan(A(k, :));
  A(k, index) = A(k - 1, index);
end

[Richard]

not homework, just needed some advice...

[Image Analyst]

But you didn't say whether you solved it or not. If you did, mark this as "Answered" otherwise show your code and where you're having difficulty.

Answer 2

[Matt J]

The FEX has this file

http://www.mathworks.com/matlabcentral/fileexchange/4551-inpaintnans

as well as many variants linked to it.