我有两个数据帧。 一旦常规数据帧：
DF
Build Artifacts和包含假日日期的Dataframe。例如，如果 DF 的日期范围是2014-2015，那么DF-Holiday将有2014年和2015年的假期：
DFHolidays
Datum
...
2014-12-30 23:00:00
2014-12-30 23:15:00
2014-12-30 23:30:00
2014-12-30 23:45:00
2014-12-31 00:00:00
...
2015-01-01 00:00:00
2015-01-02 00:00:00
2015-01-03 00:00:00
2015-01-04 00:00:00
2015-01-04 00:00:00
2015-01-05 00:00:00
...
Dataframe＆＃34; DF＆＃34;现在应该为每个假期都有一个新列，它计算给定行的每年假日的天数。
示例：
DATE
NAME
Neujahr 2014-01-01
Heilige Drei Könige 2014-01-06
Karfreitag 2014-04-18
Ostersonntag 2014-04-20
Ostermontag 2014-04-21
1. Mai 2014-05-01
...
Erster Weihnachtsfeiertag 2015-12-25
Zweiter Weihnachtsfeiertag 2015-12-26
...
我写了以下代码。
Neujahr Heilige Drei Könige Karfreitag ...
...
2014-01-01 23:00:00 0 value value
2014-01-02 23:15:00 1 value value
2014-01-03 23:30:00 2 value value
2014-01-04 23:45:00 3 value value
2014-01-05 00:00:00 4 value value
...
这有效，但速度很慢。我怎么能更优雅地解决这个问题？
编辑：完整的DFHolidays（在这个例子中只有两年。真正的数据大约是10年以上。
import datetime
from datetime import datetime
def computeDiff(x,y):
x = pd.to_datetime(x).date()
print("x: ",x)
mask = mask = (y.dt.year == x.year)
y = y.loc[mask]
y = y.get_value(0,0)
print("y: ",y)
y = pd.to_datetime(y).date()
return (y - x).days
for holiday in list(DFHolidays.index):
day = DFHolidays.loc[holiday, 'DATE']
df[holiday] = df['Datum'].apply(computeDiff, args=(day,))
答案 0 :(得分：1)
以下内容如何：
import pandas as pd
df = pd.DataFrame({'Date': ['2014-12-30 23:00:00','2014-12-30 23:15:00','2014-12-30 23:30:00','2014-12-30 23:45:00','2014-12-31 00:00:00']})
DFHolidays = pd.DataFrame({'NAME': ['Neujahr', 'Heilige Drei Könige', 'Karfreitag', 'Ostersonntag', 'Ostermontag', '1. Mai', 'Erster Weihnachtsfeiertag', 'Zweiter Weihnachtsfeiertag'],
'DATE': ['2014-01-01','2014-01-06','2014-04-18','2014-04-20','2014-04-21','2014-05-01','2015-12-25','2015-12-26']})
# Ensure all dates are actually dates
df['Date'] = pd.to_datetime(df['Date'])
DFHolidays['DATE'] = pd.to_datetime(DFHolidays['DATE'])
DFHolidays.set_index('NAME', inplace=True)
# Loop over each holiday, apply the calculation
for holiday_name, date in DFHolidays['DATE'].to_dict().items():
df[holiday_name] = date - df['Date']
返回，给出样本数据：
Date Neujahr Heilige Drei Könige \
0 2014-12-30 23:00:00 -364 days +01:00:00 -359 days +01:00:00
1 2014-12-30 23:15:00 -364 days +00:45:00 -359 days +00:45:00
2 2014-12-30 23:30:00 -364 days +00:30:00 -359 days +00:30:00
3 2014-12-30 23:45:00 -364 days +00:15:00 -359 days +00:15:00
4 2014-12-31 00:00:00 -364 days +00:00:00 -359 days +00:00:00
Karfreitag Ostersonntag Ostermontag \
0 -257 days +01:00:00 -255 days +01:00:00 -254 days +01:00:00
1 -257 days +00:45:00 -255 days +00:45:00 -254 days +00:45:00
2 -257 days +00:30:00 -255 days +00:30:00 -254 days +00:30:00
3 -257 days +00:15:00 -255 days +00:15:00 -254 days +00:15:00
4 -257 days +00:00:00 -255 days +00:00:00 -254 days +00:00:00
1. Mai Erster Weihnachtsfeiertag Zweiter Weihnachtsfeiertag
0 -244 days +01:00:00 359 days 01:00:00 360 days 01:00:00
1 -244 days +00:45:00 359 days 00:45:00 360 days 00:45:00
2 -244 days +00:30:00 359 days 00:30:00 360 days 00:30:00
3 -244 days +00:15:00 359 days 00:15:00 360 days 00:15:00
4 -244 days +00:00:00 359 days 00:00:00 360 days 00:00:00
了解数据年份非常重要，我们可以执行以下操作：
import pandas as pd
df = pd.DataFrame({'Date': ['2016-02-15 23:00:00','2016-03-05 23:15:00','2016-12-30 23:30:00','2017-08-10 23:45:00','2017-09-01 00:00:00']})
DFHolidays = pd.DataFrame({'NAME': ['Neujahr','Karfreitag','Ostersonntag', 'Ostermontag', '1. Mai','Christi Himmelfahrt','Pfingstsonntag','Pfingstmontag', 'Tag der deutschen Einheit', 'Erster Weihnachtsfeiertag', 'Zweiter Weihnachtsfeiertag','Neujahr','Karfreitag','Ostersonntag', 'Ostermontag','1. Mai','Christi Himmelfahrt','Pfingstsonntag','Pfingstmontag', 'Tag der deutschen Einheit', 'Erster Weihnachtsfeiertag', 'Zweiter Weihnachtsfeiertag'],
'DATE': ['2016-01-01','2016-03-25','2016-03-27','2016-03-28','2016-05-01','2016-05-05','2016-05-15','2016-05-16','2016-10-03','2016-12-25','2016-12-26','2017-01-01','2017-04-14','2017-04-16','2017-04-17','2017-05-01','2017-05-25','2017-06-04','2017-06-05','2017-10-03','2017-12-25','2017-12-26']})
# Ensure all dates are actually dates
df['Date'] = pd.to_datetime(df['Date'])
DFHolidays['DATE'] = pd.to_datetime(DFHolidays['DATE'])
# Set up a year column in both dataframes ot join on shortly
DFHolidays['Year'] = DFHolidays['DATE'].dt.year
df['Year'] = df['Date'].dt.year
# Work out what all the holiday names are
holiday_names = DFHolidays['NAME'].unique()
DFHolidays = DFHolidays.pivot(index='Year', columns='NAME', values='DATE') \
.reset_index()
# Merge the frames
df = df.merge(DFHolidays, on='Year')
# Calculate the difference
for holiday in holiday_names:
df[holiday] = df[holiday] - df['Date']
这给了我们：
Date Year 1. Mai Christi Himmelfahrt \
0 2016-02-15 23:00:00 2016 75 days 01:00:00 79 days 01:00:00
1 2016-03-05 23:15:00 2016 56 days 00:45:00 60 days 00:45:00
2 2016-12-30 23:30:00 2016 -244 days +00:30:00 -240 days +00:30:00
3 2017-08-10 23:45:00 2017 -102 days +00:15:00 -78 days +00:15:00
4 2017-09-01 00:00:00 2017 -123 days +00:00:00 -99 days +00:00:00
Erster Weihnachtsfeiertag Karfreitag Neujahr \
0 313 days 01:00:00 38 days 01:00:00 -46 days +01:00:00
1 294 days 00:45:00 19 days 00:45:00 -65 days +00:45:00
2 -6 days +00:30:00 -281 days +00:30:00 -365 days +00:30:00
3 136 days 00:15:00 -119 days +00:15:00 -222 days +00:15:00
4 115 days 00:00:00 -140 days +00:00:00 -243 days +00:00:00
Ostermontag Ostersonntag Pfingstmontag \
0 41 days 01:00:00 40 days 01:00:00 90 days 01:00:00
1 22 days 00:45:00 21 days 00:45:00 71 days 00:45:00
2 -278 days +00:30:00 -279 days +00:30:00 -229 days +00:30:00
3 -116 days +00:15:00 -117 days +00:15:00 -67 days +00:15:00
4 -137 days +00:00:00 -138 days +00:00:00 -88 days +00:00:00
Pfingstsonntag Tag der deutschen Einheit Zweiter Weihnachtsfeiertag
0 89 days 01:00:00 230 days 01:00:00 314 days 01:00:00
1 70 days 00:45:00 211 days 00:45:00 295 days 00:45:00
2 -230 days +00:30:00 -89 days +00:30:00 -5 days +00:30:00
3 -68 days +00:15:00 53 days 00:15:00 137 days 00:15:00
4 -89 days +00:00:00 32 days 00:00:00 116 days 00:00:00
要获得天数，请将最后一行更改为
df[holiday] = df[holiday].dt.date - df['Date'].dt.date