How to convert 'YYYY-mm-dd HH:MM:SS.[]' to number of minutes in the date?

Question

I have a dataframe df with column date in format 'YYYY-mm-dd HH:MM:SS.miliseconds' (datetime64[ns]). I want to generate another column, minutes (float), as the number of minutes counted from time 00:00:00 to time HH:MM:SS. How can I do that?

Answer 1

you can easily convert to any unit of time since midnight by

• subtracting the date from the datetime to get a timedelta (duration),
• then convert that timedelta to the unit you require;

EX:

import pandas as pd

df = pd.DataFrame({"datetime": ["2016-01-01 00:00:00",
                                "2016-03-01 12:00:00",
                                "2016-06-01 23:59:00"]})

df["datetime"] = pd.to_datetime(df["datetime"])

# subtract the datetime floored to the day to get a timedelta,
df["minutes_from_midnight"] = (
    (df["datetime"] - df["datetime"].dt.floor("d"))
    .dt.total_seconds() # then convert to minutes
    .div(60)
)

df
             datetime  minutes_from_midnight
0 2016-01-01 00:00:00                    0.0
1 2016-03-01 12:00:00                  720.0
2 2016-06-01 23:59:00                 1439.0

Answer 2

I used the following solution. First, I filtered the hours, minutes, and seconds from the timestamp (column date):

df['time_hours'] = df.date.dt.hour
df['time_minutes'] = df.date.dt.minute
df['time_seconds'] = df.date.dt.second

Then, I defined a function to calculate total time in minutes, using simple algebra:

def get_total_minutes(row):
    time_hours = row['time_hours']
    time_minutes = row['time_minutes']
    time_seconds = row['time_seconds']
    total_minutes = float((time_hours*60)+(time_minutes)+(time_seconds/60))
    return total_minutes

Finally, I applied the function to each row and got the intended results:

df['minutes'] = df.apply(lambda row: get_total_minutes(row), axis=1)