Using the new datetime structure within datasets

I have a dataset with date variables as datenums. I'd like to convert them to the new datetime datatype. Here is what I did:

ds.newdate = datetime(ds.olddate,'ConvertFrom','datenum');

This is what I get:

ds
ans =
olddate      newdate
734138       [1x1 datetime]

This isn't useful. What I want is this:

    ds
    ans =
    olddate      newdate
    734138       31-Dec-2009

There is some sort of sub-indexing going on where I have to go another level deeper to access my date variable. How do I access it directly in the dataset?

Answer 1

[Peter Perkins]

Jesse, have you tried using a table rather than a dataset? A table is, roughly speaking, a replacement for dataset that is available in core MATLAB (rather than just the Statistics Toolbox) since R2013b. Since you are using datetime, you must have R2014b.

The [1x1 datetime] that you are seeing is really just a display artifact, and ds.newdate is a datetime. But table provides the display you are looking for:

>> t = table((734138:734140)','VariableNames',{'OldDate'})
t = 
     OldDate  
    __________
    7.3414e+05
    7.3414e+05
    7.3414e+05
>> t.NewDate = datetime(t.OldDate,'ConvertFrom','datenum')
t = 
     OldDate            NewDate       
    __________    ____________________
    7.3414e+05    31-Dec-2009 00:00:00
    7.3414e+05    01-Jan-2010 00:00:00
    7.3414e+05    02-Jan-2010 00:00:00

Hope this helps.

[Jesse Blocher]

Great! I didn't even know about tables. I had been using datatables for a while and must have missed that new structure.