CODEWITHSUNDEEP
matlabdatevector
Constantin
Constantin

Reputation: 669

How to ensure that a vector of date numbers is fully contained in another in MATLAB

I have two vectors A and B containing date numbers. In theory, all coefficients in B should also be contained in A. Unfortunately this is not the case. My theory justifies replacing all non-corresponding date numbers with the next date number in the A vector, for which I would like to write a script.

More concretely: Both vectors contain dates of financial transactions. Vector A contains all trading days of the time period examined so that all of the transaction dates in B should also be contained in A because transactions cannot take place on non-trading days. Where an entry in B is not in A I hypothesise that trades were mistakenly reported to have taken place on a non-trading day and I would thus like to replace these dates with the next highest number that is contained in A.

To provide an example, I would like to turn this:

A = [2;3;4;7]
B = [2;3;4;5;6;7]

Into this:

A = [2;3;4;7]
B = [2;3;4;7;7;7]

So that B(4:5) were modified because they are not contained in A. The script would have to identify that 5 is not contained in A and it would then check whether 5+i is contained in A until this evaluates to true (which is the case at 7). At this point the coefficient would be inserted into B to replace the old date.

Note that this question is closely related to this earlier question; however it is not identifying business days that this question is about. Instead I want to accept one of my date vectors (A) as authoritative (i.e. adequately representing trading days including national holidays) and base the substitution of the non-conforming entries on the A vector instead of a Bloomberg function.

Upvotes: 2

Views: 37

Answers (2)

Luis Mendo
Luis Mendo

Reputation: 112699

You can apply bsxfun to compare each element of A with each element of B, and then use the second output of max to find the required indices:

[~, ind] = max(bsxfun(@le, B(:).', A(:)), [], 1);
B_result = A(ind);

This assumes that for each element of B there is at least one element in A which is greater or equal. Values need not be integer or sorted. For example,

A = [2; 3.5; 4.5; 7; 10];
B = [0.1; 2; 2; 6; 5; 4.5];

gives

B_result =
    2.0000
    2.0000
    2.0000
    7.0000
    7.0000
    4.5000

Upvotes: 1

Jeff Irwin
Jeff Irwin

Reputation: 1051

This solution assumes that a and b are already sorted. If not, then simply replace them with sort(a) and sort(b).

a = [ 2; 3; 4; 7 ];
b = [ 2; 3; 4; 5; 6; 7 ];

% Check each element in b
for k = 1: length(b)

    % If this element of b is not contained in a
    if (all(a(:) ~= b(k)))

        % Replace with the next element in a that is greater than b(k)
        b(k) = a(find(a(:) > b(k), 1, 'first'));
    end
end

Upvotes: 2

Related Questions