Minutes off in datetime formatting

[Milla Hanzon]

Hello,

I am working on a load profile with 30-minute invervals of electricity demand that I wish to collapse into hourly demand in Stata. In order to do so, I have converted a string variable to a date-time format using the clock function in Stata:


gen DateTime2 = clock(DateTime, "20YMDhm")

format %tcMon_dd,_CCYY_HH:MM DateTime2

The problem is that the minutes are slightly off. DateTime in the list below is the string variable and DateTime2 is the converted variable. As you can see, 1.30 becomes 1.29 and so on.


list DateTime in 1/10

+------------------+
| DateTime |
|------------------|
1. | 2018-07-01 00:30 |
2. | 2018-07-01 01:00 |
3. | 2018-07-01 01:30 |
4. | 2018-07-01 02:00 |
5. | 2018-07-01 02:30 |
|------------------|
6. | 2018-07-01 03:00 |
7. | 2018-07-01 03:30 |
8. | 2018-07-01 04:00 |
9. | 2018-07-01 04:30 |
10. | 2018-07-01 05:00 |
+------------------+

. list DateTime2 in 1/10

+-------------------+
| DateTime2 |
|-------------------|
1. | Jul 1, 2018 00:30 |
2. | Jul 1, 2018 01:00 |
3. | Jul 1, 2018 01:29 |
4. | Jul 1, 2018 01:59 |
5. | Jul 1, 2018 02:30 |
|-------------------|
6. | Jul 1, 2018 03:00 |
7. | Jul 1, 2018 03:29 |
8. | Jul 1, 2018 03:59 |
9. | Jul 1, 2018 04:30 |
10. | Jul 1, 2018 05:01 |
+-------------------+


I wish to collapse the data to hourly demand instead of half-hourly, but I need to correct this error first. Any ideas on how to do so are highly appreciated!

Milla

[Nick Cox]

help datetime

tells users again and again that clock variables must be created double to avoid precision problems.