find multiple words in a cell

Hi, is there a way to compare and return a number for the matching words in two unequal cells?

A={'a','a','a','b','b','d'} B={'a','b','c','d','e','f','g','h'}

return C=3,2,0,1,0,0,0,0

Thank you

Answer 1

[Ryan Livingston]

A one-liner:

cellfun(@(x)sum(ismember(A,x)), B)

[Vincent I]

Fantastic, thank you . thats what I was looking for

Answer 2

[Matt J]

If the "words" will really always be single letters, you could do it looplessly with

>> histc([A{:}], [B{:}])

ans =

       3     2     0     1     0     0     0     0

[Vincent I]

ok I missed something on my initial comment. what if

A={'a_1','ab','a','b','b','d'}

B={'a','b','c','d','e','f','g','h'}

return C=3,2,0,1,0,0,0,0

[Matt J]

Post this as a new question. When you do, explain why 'b' gets only a count of 2 even though it occurs 3 times in A. Also explain what would happen if a letter occurs twice in the same string, e.g.

A={'a_1ab','bb'}; B={'a','b'};

[Vincent I]

ohh I see... b should get a count of two...

A={'a_1','a_2','a','b','b','d'}

B={'a','b','c','d','e','f','g','h'}

return C=3,2,0,1,0,0,0,0

And, although this is a new question, I feel that it is not far away for the initial question which I dont see why should require a new post

thank you

[Daniel Shub]

@Vincent So you are looking for non-exact matches? I don't see your accepted answer works for that.

[Matt J]

And, although this is a new question, I feel that it is not far away for the initial question which I dont see why should require a new post

It doesn't require it, but if you post it as a new question, people have the opportunity to gain points from answering you (now that you've already accepted Ryan's answer), and so will be more incentivized to do so.

[Jan]

The decision is easy:

 New question, new thread.

And:

 Additional information to an existing question is added by editing the question and marking the changes by "[EDITED]". Then this is clarification and *not* a new question.

Hiding important information in deeply nested comments to already accepted questions is a bad idea.

Answer 3

[Daniel Shub]

I am sure that this is over thinking the solution and I doubt that using regexp is optimal, but I was curious how bad it would be.

x = regexp(A, cell2mat(cellfun(@(x)['(?<', x, '>', x, ')|'], B, 'UniformOutput', false)), 'names');
cellfun(@(x)length([y.(x)]), fieldnames([x{:}]))'

I is there a better way to do this with regexp?

[Vincent I]

Solved my problem by doing the folowing:

A=TF(:,1).';

A=regexprep(A, ' ','');

A=regexprep(A, '_(\w*)','');

B={'aa','bb2c','bb25c','xy3c','m56c','etc56c'};

C=cellfun(@(x)sum(strcmp(A, x)), B);

Thank you

Answer 4

[Jan]

[EDITED, Jan, moved from comments to the accepted question]

ISMEMBER sorts the inputs and performs a binary search. This can be much faster and much slower than an unsorted comparison by:

cellfun(@(x)sum(strcmp(A, x)), B)

I claim without a proof, that a loop is faster:

R = zeros(1, numel(B));
for iB = 1:numel(B)
  R(iB) = sum(strcmp(A, B{iB}));
end

[EDITED 2] And if you want to compare the leading character(s) only:

...
  R(iB) = sum(strncmp(A, B{iB}, length(B{iB}));
...

[Vincent I]

sounds goood. Thanks for your answer