Change day start and end time - r

Question

How can I force R to change the parameters of a day. So I mean that for example yrDay like here provided doesn't go from 0am til 0am but from 6pm til 6pm.
df <- data.frame(Date=seq(
from=as.POSIXct("2012-1-1 13:00:00", tz="UTC"),
to=as.POSIXct("2012-1-3 13:00:00", tz="UTC"),
by="hour")
)
df$yrDay <- as.numeric(strftime(df$Date,format="%j"))

Answer 1

Just add 5 hours (5 * 60 min * 60 sec) using your code:
df$yrDay <- as.numeric(strftime(df$Date + 5*60*60,format="%j"))
Date yrDay
1 2012-01-01 13:00:00 1
2 2012-01-01 14:00:00 1
3 2012-01-01 15:00:00 1
4 2012-01-01 16:00:00 1
5 2012-01-01 17:00:00 1
6 2012-01-01 18:00:00 2
7 2012-01-01 19:00:00 2
8 2012-01-01 20:00:00 2
9 2012-01-01 21:00:00 2
10 2012-01-01 22:00:00 2
...
Or 6 hours using lubridate (better approach IMHO):
df$yrDay <- lubridate::day(df$Date + 6*60*60)
But as mentioned by #ngm in the comments, it is a quick and dirty solution that might not be robust for all cases.

Answer 2

A robust solution in python code for the Q above.
This algorithm uses a NumPy and pandas to set a conditions for each time in with your datetime data. It simply ask if the day before is less then the time specified and the is date == to the current date.
Then it ask and if the day after is greater then shift hour and is it This condition been generate for each day in your data then is check for date same as the before day.
Hope this help or encourage other to try think for better solution.
Sorry about my English it is not my first and I am very busy so this is my best for the time i allocate to make an answer. Hope to improve please give helpful comments and try not to harsh :)
def count_custome_day_start_and_end(pd_index_datetime, shift_time='8:30am'):
dates = sorted(set(pd_index_datetime.date))
times = np.array([time.time() for time in pd_index_datetime
])
shift_time = pd.to_datetime(shift_time).time()
conditiones = [
cond for date in dates for cond in [
((times < shift_time)
& (pd_index_datetime.date == date)), # day before
((times >= shift_time)
& (pd_index_datetime.date == date)) # day after
]
]
langth = (len(conditiones) // 2) + 1
choices = [0, *np.repeat(range(1, langth), 2)]
choices.pop()
return np.select(conditiones, choices, pd_index_datetime.factorize()[0])