how to reshape data with leap year?

I have 30 years daily data. For reason, i want to reshape the data as days*years (e.g. 365/366, 30). for the year with no leap day it would be easy (reshape(Data,365,[]). how one would do it with leap day included?. see the code I tried but is not working. the data I have is from 1981-2010.

Data=(1:10957)';
R=reshape(Data,[],30); % of course it wont work because of the dimensions

[KL]

which version are you using?

[Guillaume]

By definitions all columns of a matrix have the same number of rows. You can't have rows with 365 elements and others with 366 elements.

So what do you want to do? Fill up the last row of non-leap year with NaN? Store as a 1x30 cell array of 365x1 or 366x1 vectors?

Both will most likely make it harder to process your data than as it is now.

[Hydro]

I just want the yearly average of the data. I thought it would be easy to reshape the daily data into (daily*yearly) and then just take the average. KL! I am using 2017 version.

[Walter Roberson]

Then there is the old "all entries with the same calendar date must appear in the same row", in which case you get NaN in the middle for each Feb 29 that is not present.

Answer 1

[KL]

No need to reshape. Use timetables.

https://de.mathworks.com/help/matlab/matlab_prog/create-timetables.html

https://de.mathworks.com/help/matlab/ref/retime.html

then you can simply use

 yearlymeantable = retime(yourtimetable,'yearly','mean')

[Hydro]

WOW.. that is an awesome function. i tried the following, however, it gives means for January only?

t1=(datetime(1981,1,1):datetime(2010,12,31))';
TT=timetable(t1,Temp);
TT2=retime(TT,'yearly','mean');

[Walter Roberson]

No, it is producing an average for each year. It has to pick a representative date from each year to associate the average with, because MATLAB does not have a time object that has "year only". If you want just the year to show up, you could use

TT2.t1.Format = 'yyyy';

[Hydro]

Gotcha. Thanks