MATLAB: Get new Double Matrix from Data on a Cell Array

cell arraycountMATLABnew variable

I have a cell variable with 13000 rows and 11 columns. For instance:
     C1 c2      c3      c4      c5      c6      c7
  A={22  1988  1992  2001  2001  []  []
     23  1983  1988  []  []  []  []
     29  1983  1985  1988  1991  2000  2003}
C1 is a code for an individual; c2 is the year the individual first is in the sample; c3 is the year he/she leaves. C4 is the year the individual comes back to the sample; c5 is the year the individual goes off again; and so on. The sample period is from 1983 to 2013. I can have individuals that appear in one year, leave in other and never appear again; as others may be constantly on and off the sample. I am trying to get the following variable:
  • A double variable with 3 rows, ordered by year, indicating for each year and for each individual the number of years the individual is in the sample in year x.
  • If the indiviual leaves the period for at least one year and comes back again , I start the count from one (number of years the individual is in the sample).
  • If the individual is less than 3 years in the sample don't consider it for the count .
B=[1983 23 1 
1983 29 1
1984 23 2
1984 29 2
1985 23 3
1985 29 3
1986 23 4
1987 23 5
1988 22 1
1988 23 6
1988 29 1
1989 22 2
1989 29 2
1990 22 3 
1990 29 3
1991 22 4
1991 29 4
1992 22 5
…
2001 29 2
2002 29 3
2003 29 4]
Can someone help me? Thank you.

Best Answer

Something like this would do:
B = [];
for row = 1:size(A, 1)
    personid=A{row, 1};
    for col = 2:2:size(A, 2)
        startyear = A{row, col};
        endyear = A{row, col+1};
        if endyear - startyear >= 2 %also works if endyear and startyear are empty
            for year = startyear:endyear
                B = [B; year personid year-startyear+1];
            end
        end
    end
end
B = sortrows(B, [1 2]);
This assumes that there's always a leaving year.
The only issue with this code is the constant resizing of B. You could preallocate a huge enough array (it can't have more rows than number of persons * 30 years) and resize at the end to the actual size if it's a problem for you
Related Question