This will be much easier if you convert the month, day columns into date numbers. I don't have the newer table by which you could then use the new datetime and splitapply, but the idea can be easily illustrated for just plain arrays...
>> y=[2005 2005 2006 2006 2006];
>> m=[12 12 1:3];
>> dn=datenum(y,m,1).'; % convert to date numbers
>> u=unique(dn); % the unique dates in the total list
>> n=hist(dn,u) % bin over those values
n =
2 1 1 1
>> ds=dataset(u,datestr(u,'mmm yyyy'),n.','VarNames',{'DateNum';'Date';'Events'})
ds =
DateNum Date Events
7.3265e+05 Dec 2005 2
7.3268e+05 Jan 2006 1
7.3271e+05 Feb 2006 1
7.3274e+05 Mar 2006 1
>>
Last simply uses the Statistics Toolbox dataset to display the disparate values similar to what the new table class can do...
