problem with dateshift..how solve it?

data1=datetime("2023-01-01")

data1 = datetime
   01-Jan-2023

data2=datetime("2023-11-15")

data2 = datetime
   15-Nov-2023

dt=data1:days(7):data2

dt = 1×46 datetime array
Columns 1 through 14

   01-Jan-2023   08-Jan-2023   15-Jan-2023   22-Jan-2023   29-Jan-2023   05-Feb-2023   12-Feb-2023   19-Feb-2023   26-Feb-2023   05-Mar-2023   12-Mar-2023   19-Mar-2023   26-Mar-2023   02-Apr-2023

Columns 15 through 28

   09-Apr-2023   16-Apr-2023   23-Apr-2023   30-Apr-2023   07-May-2023   14-May-2023   21-May-2023   28-May-2023   04-Jun-2023   11-Jun-2023   18-Jun-2023   25-Jun-2023   02-Jul-2023   09-Jul-2023

Columns 29 through 42

   16-Jul-2023   23-Jul-2023   30-Jul-2023   06-Aug-2023   13-Aug-2023   20-Aug-2023   27-Aug-2023   03-Sep-2023   10-Sep-2023   17-Sep-2023   24-Sep-2023   01-Oct-2023   08-Oct-2023   15-Oct-2023

Columns 43 through 46

   22-Oct-2023   29-Oct-2023   05-Nov-2023   12-Nov-2023

   RP=dateshift(data1:calmonths(1):data2, 'end', 'month');  
   RP = dateshift(RP,'dayofweek','Saturday','previous');
   RP(end)

ans = datetime
   25-Nov-2023

why RP(end) is > data2 ?

i write: dt=data1:days(7):data2 the result must to be <=data2 but it's > data2

Answer 1

[Walter Roberson]

dt=data1:days(7):data2 does end before data2 -- the last entry is the 12-Nov-2023 that you see displayed.

But then you take something that starts at the beginning of the month of January 2023, and advance one calendar month at a time ending no later than data2 . The beginning of Novemeber 2023 is several calendar months after January 1 2023, so the data1:calmonths(1):data2 series ends with a November 1 2023 datetime object. you then dateshift() that to the end of the month, getting a November 30, 2023 datetime object. You then dateshift() that to the previous Saturday, which gets you the November 25 2023 datetime object. There is no reason why the last Saturday of a month must be before a datetime that is mid-month.

[Luca Re]

How do I do the calculation but it must be within data2(end)?

[Walter Roberson]

You have given code, and expressed that the code does not do what you want, but you have not expressed what you do want?

Are you looking for the last saturday that is no later than data2 ?

Are you looking for the last saturday that is no later than dt(end) ?

Are you looking for the last saturday of a month that is no later than data2 ? If so then

dateshift(dateshift(data2, 'start', 'month') - 1, 'dayofweek', 'saturday', 'previous')

... but what do you want to do about the case where data2 is in the same calendar month as data1?

Hmmm... that dateshift line I gave is possibly wrong for the case where data2 is the last saturday of a month -- but whether it is wrong or not depends on your intention with the days(7) and calmonths(1) . For example if data1 is the monday of the last full week of November 2023, and data2 is the saturday of that week exactly, then the description "last Saturday of a month that is at least data1 and is no more than data2" would suggest data2 as the result -- but with your days(7) and calmonths(1) iterations, the only entry in the vectors would be data1 and applying the logic to that would lead you to the last saturday in October 2023 which would be before data1 in this case...

So you need to define clearly what exactly you are searching for.

[Luca Re]

i try the solution as:

RP(trova(RP<dt(fine)))

it's correct

thank you

[Walter Roberson]

I suspect you meant

RP(find(RP<dt(end)))

[Luca Re]

😁👍 I had "translate" turned on

[Stephen23]

Most likely FIND is not required, logical indexing is simpler and more efficient:

RP(RP<dt(end))

[Luca Re]

thank

Answer 2

[Mann Baidi]

Hi Luca,;

I understnad you are facing issues in getting the desired output using the "dateshift" function. You are encountering the above issue because when the code is executed, the array "RP" has the end value as "30-Nov-2023" after the execution of line 4.

  
data1=datetime("2023-01-01");
data2=datetime("2023-11-15");
dt=data1:days(7):data2;
RP=dateshift(data1:calmonths(1):data2, 'end', 'month')

RP = 1×11 datetime array
   31-Jan-2023   28-Feb-2023   31-Mar-2023   30-Apr-2023   31-May-2023   30-Jun-2023   31-Jul-2023   31-Aug-2023   30-Sep-2023   31-Oct-2023   30-Nov-2023

RP = dateshift(RP,'dayofweek','Saturday','previous');
RP(end);

This is the reason "RP(end)" is giving "25-Nov-2023" as it is the previous saturday for "30-Nov-2023".

However, if you would like to avoid this issue and would like to find the previous Saturday of "data2" you can modify the code by passing "RP(end-1)" instead of "RP(end)" in line 5 of the code., and then check for the "data2"separately. Here is the possible workaround.

data1=datetime("2023-01-01");
data2=datetime("2023-11-15");


dt=data1:days(7):data2;


   RP=dateshift(data1:calmonths(1):data2, 'end', 'month');  
   RP = dateshift(RP(end-1),'dayofweek','Saturday','previous');
   RP(end)

ans = datetime
   28-Oct-2023

lastSat= dateshift(data2,'dayofweek','Saturday','previous')

lastSat = datetime
   11-Nov-2023

Hope this will resolve your doubt and query!

[Luca Re]

thank but i thik i code this:

RP(find(RP<dt(end)))

ans =

1×10 datetime array

Columns 1 through 7

2023-01-28 2023-02-25 2023-03-25 2023-04-29 2023-05-27 2023-06-24 2023-07-29

Columns 8 through 10

2023-08-26 2023-09-30 2023-10-28