count of months spanned

I can't think of a way to do this without a loop. I have a bunch of start and end dates. Each row is a record and I want to be about to count how many months are in each date range. Here's some sample data startdates=[735406;735416;735404;735396;735363;735389]; enddates=[735433;735433;735433;735425;735416;735416];

So Jun 21-Jul 18 would be two months

>>countmonths =

     2     1     2     2     3     2

My way is not very efficient

for i=1:length(startdates)
daterange=startdates(i):enddates(i);
dv=datevec(daterange);
countmonths(i)=size(unique(dv(:,[1 2]),'rows'),1);
end

Answer 1

[Sven]

Hi Leah,

I think you can do this quite nicely as follows:

startvecs = datevec(startdates);
endvecs = datevec(enddates);
diffvecs = endvecs - startvecs;
countmonths = 12*diffvecs(:,1) + diffvecs(:,2)+1

Does that work for you? I'm not quite sure what you'd want to do between, say, Jan1 and Feb1, but I think if you take a look at diffvecs you'll understand what it contains and be able to make it work for your purposes.

[Leah]

Thanks, this does work for me. Jan1 and Feb1 would still count as two months, which is correct for my purposes. Thanks!

Answer 2

[Jan]

When you convert the serial date numbers to date vectors you get e.g.:

s = [2013 06 21 12 13 14; ...
     2013 06 21  1  2  3]
e = [2013 07 18 12 13 14; ...
     2014 06 21  1  2  3]

Now the distance in months is:

(e(:, 1) * 12 + e(:, 2)) - (s(:, 1) * 12 + s(:, 2)) + 1

Or more compact:

(e(:, 1:2) - s(:, 1:2)) * [12; 1] + 1

[Leah]

Thanks Jan! This was really similar to the first post they both work great