Idea is floor datetimes for remove times
by floor by days and get number of business days between start day + one day to hours3
column by numpy.busday_count
and then create hour1
and hour2
columns for start and end hours with floor by hours if not weekends hours. Last sum all hours columns together:
df = pd.DataFrame(columns=['Inflow_date_time','End_date_time', 'need'])
df.Inflow_date_time= [pd.Timestamp('2019-08-01 23:22:46'),
pd.Timestamp('2019-08-03 23:22:46'),
pd.Timestamp('2019-08-01 23:22:46'),
pd.Timestamp('2019-07-26 23:22:46'),
pd.Timestamp('2019-08-05 11:22:46')]
df.End_date_time= [pd.Timestamp('2019-08-05 17:43:51')] * 5
df.need = [42,17,41,138,6]
#print (df)
df["hours1"] = df["Inflow_date_time"].dt.ceil('d')
df["hours2"] = df["End_date_time"].dt.floor('d')
one_day_mask = df["Inflow_date_time"].dt.floor('d') == df["hours2"]
df['hours3'] = [np.busday_count(b,a)*24 for a, b in zip(df['hours2'].dt.strftime('%Y-%m-%d'),
df['hours1'].dt.strftime('%Y-%m-%d'))]
mask1 = df['hours1'].dt.dayofweek < 5
hours1 = df['hours1'] - df['Inflow_date_time'].dt.floor('H')
df['hours1'] = np.where(mask1, hours1, np.nan) / np.timedelta64(1 ,'h')
mask2 = df['hours2'].dt.dayofweek < 5
df['hours2'] = (np.where(mask2, df['End_date_time'].dt.floor('H')-df['hours2'], np.nan) /
np.timedelta64(1 ,'h'))
df['date_diff'] = df['hours1'].fillna(0) + df['hours2'].fillna(0) + df['hours3']
one_day = (df['End_date_time'].dt.floor('H') - df['Inflow_date_time'].dt.floor('H')) /
np.timedelta64(1 ,'h')
df["date_diff"] = df["date_diff"].mask(one_day_mask, one_day)
print (df)
Inflow_date_time End_date_time need hours1 hours2 hours3
0 2019-08-01 23:22:46 2019-08-05 17:43:51 42 1.0 17.0 24
1 2019-08-03 23:22:46 2019-08-05 17:43:51 17 NaN 17.0 0
2 2019-08-01 23:22:46 2019-08-05 17:43:51 41 1.0 17.0 24
3 2019-07-26 23:22:46 2019-08-05 17:43:51 138 NaN 17.0 120
4 2019-08-05 11:22:46 2019-08-05 17:43:51 6 13.0 17.0 -24
date_diff
0 42.0
1 17.0
2 42.0
3 137.0
4 6.0
与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…