Only month and year matrix

Hello! Can anybody help me with a datenum problem? I want to make an only month and year matrix, but I dont find a way to do this with datenum function, because makes me put the day. My dataset just have month, year and temperature. I tried to make a matrix of mm,yyy and dd and later make an bucle that gives me just one day of each month for each year (1946-2003), but then, when I make a matrix of date and temperature, matlab gives me the matrix with cientific notation.

Pd. When I put datenum(year,month,'mm/yy'), I got an error with datenum.

Answer 1

[dpb]

Use the facility of datenum to automagically roll over the higher-granularity fields...

dn=datenum(1946,[1:((2003-1946)+1)*12].',1);

It's simple for the fields that are fixed number of units such as months as every year has twelve months; for others one has to sometimes compute the number of days in the year or month,year, but the same idea works.

>> dn=datenum(1946,[1:((2003-1946)+1)*12].',1);
>> [y,m]=datevec([dn(1:3);dn(end-5:end)]);
>> [m y]
ans =
         1        1946
         2        1946
         3        1946
         7        2003
         8        2003
         9        2003
        10        2003
        11        2003
        12        2003
>>

I'm not sure if the new date datatype has a better way to generate vectors of dates from first/last dates or not; it might have some better facilities.

[dpb]

OK, looked at R2014b+ datetime; it doesn't include any way to do periodic dates automagically, either; thought it might. What it does do, however, is have a display feature that values are displayed as dates rather than date numbers which is the root cause of the above complaint re: "scientific notation". The above solution is, of course, also datenum so the values of dn are doubles of some large integer value; you manipulate them for display via datestr or numerically knowing what they represent or as above you can generate the corresponding date vectors.