Convert cell to matrix with mixed data types

I am trying to accomplish something relatively simple (most likely) but am struggling. I have a large cell array that I'd like to convert to a matrix. My cell array contains mostly numbers but a few '#N/A' where my source data was blank. I'd like to convert this cell array to a matrix and have those N/A's be converted to NaN. How can I go about doing this?

Thanks, Brian

Answer 1

[Andrei Bobrov]

eg

A = {[3]    [4]    [      4]     '#N/A'
    [2]    [1]    '#N/A'       [      3]
    [4]    [2]    [      5]    [      3]
    [1]    [4]    [      4]    [      2]}

A(cellfun(@ischar,A)) = {NaN}
out = cell2mat(A);

A - cell array - column

i1 = cellfun(@ischar,A)
sz = cellfun('size',A(~i1),2)
A(i1) = {nan(1,sz(1))}
C = cell2mat(A)

[Brian]

Simple, and does exactly what I need it do.

Thanks much,

Brian

Answer 2

[Sean de Wolski]

Do the sizes all line up?

You could first run your cell through a for-loop that identifies, '#N/A's and then replaces them. E.g:

C = {magic(3);1:3;'#N/A';pi*ones(1,3);'#N/A'};

for ii = numel(C):-1:1
    idx(ii) = isequal(C{ii},'#N/A');
end
C(idx) = {nan(1,3)};

C2 = cell2mat(C)