How to find the first date of each month in a datetime array?

I have a datetime array as below, and I like to find the first date of each month in the array.

d = [    '14-Jan-2018'
    '19-Jan-2018'
    '22-Jan-2018'
    '26-Jan-2018'
    '04-Feb-2018'
    '25-Feb-2018'
    '09-Apr-2018'
    '14-Apr-2018'
    '20-Apr-2018'
    '10-May-2018'
    '21-May-2018'
    '01-Jun-2018'
    '07-Jun-2018'
    '11-Jun-2018'
    '11-Jul-2018'
    '31-Jul-2018'
    '05-Sep-2018'
    '07-Sep-2018'
    '08-Sep-2018'
    '28-Sep-2018'
    '29-Sep-2018'
    '29-Oct-2018'
    '07-Nov-2018'
    '12-Nov-2018'
    '21-Nov-2018'
    '21-Nov-2018'
    '03-Dec-2018'
    '12-Dec-2018'
    '03-Jan-2019'
    '04-Jan-2019']

d = datetime(d)

The desired output is as below.

['14-Jan-2018'
'04-Feb-2018' 
'09-Apr-2018'
'10-May-2018'
'01-Jun-2018'
'11-Jul-2018'
'05-Sep-2018'
'29-Oct-2018'
'07-Nov-2018'
'03-Dec-2018'
'03-Jan-2019'
]

At first, it seems easy, but only few moments later I realized that it is a way harder and it takes me days on this problem without progress.

Please hlep with this, and thank you so much in advance for any help.

Answer 1

[dpb]

firstDate=groupsummary(d,findgroups(year(d),month(d)),@min);

gives

>> firstDate
firstDate = 
  11×1 datetime array
   14-Jan-2018
   04-Feb-2018
   09-Apr-2018
   10-May-2018
   01-Jun-2018
   11-Jul-2018
   05-Sep-2018
   29-Oct-2018
   07-Nov-2018
   03-Dec-2018
   03-Jan-2019
>> 

[Doheon Lee]

Thank you so much. It works perfectly!

[dpb]

Indeed. It often takes some time to realize the right way to go at some operations in MATLAB. The use of grouping variables here is the key instead of trying to compute differences individually from component pieces which I gather is probably what you had tried...

[Doheon Lee]

yes, it is pretty hard to realize the right way to do. What I learned from you will solve many problems that I definitely encounter in the future. Once again, thank you so much for the help :)

Answer 2

[Campion Loong]

Alternatively...

% dateshift all dates to beginning of the month, and find the indices of the first unique entry
[~,idx_first_month_date] = unique(dateshift(d,'start','month'));

% use the indices on your original datetime array
d(idx_first_month_date)

This yields the same result on your input:

ans = 
  11×1 datetime array
  
['14-Jan-2018'
'04-Feb-2018' 
'09-Apr-2018'
'10-May-2018'
'01-Jun-2018'
'11-Jul-2018'
'05-Sep-2018'
'29-Oct-2018'
'07-Nov-2018'
'03-Dec-2018'
'03-Jan-2019'
]