我有一个如下所示的熊猫数据框

start_time                     end_time   value
2017-01-09 21:49:55 2017-01-09 21:55:41  150.0
2017-01-09 21:55:41 2017-01-09 21:58:46    4.0
2017-01-09 22:00:55 2017-01-09 23:13:00  144.0


我想要

start_time                     end_time   value
2017-01-09 21:49:55 2017-01-09 21:58:46  154.0
2017-01-09 22:00:55 2017-01-09 23:13:00  144.0


由于前两行是连续事件,因此我将其合并为一行并添加了它们的值。
任何建议,我该如何进行。

最佳答案

假设您的数据框已经按时间排序

from datetime import datetime
import pandas as pd

'''
start_time                     end_time   value
2017-01-09 21:49:55 2017-01-09 21:55:41  150.0
2017-01-09 21:55:41 2017-01-09 21:58:46    4.0
2017-01-09 22:00:55 2017-01-09 23:13:00  144.0
'''

# your dataframe

df = pd.DataFrame({'start_time': [datetime(2017,1,9,21,49,55), datetime(2017,1,9,21,55,41),datetime(2017,1,9,22,00,55)], \
                    'end_time': [datetime(2017,1,9,21,55,41), datetime(2017,1,9,21,58,46),datetime(2017,1,9,23,13,00)], \
                    'value': [150.0, 4.0, 144.0]})


获取连续时间范围的第一个start_time和连续时间范围的最新end_time

df['start_time_'] = df['start_time'].loc[df['end_time'].shift(1) != df['start_time']]
df['end_time_'] = df['end_time'].loc[df['end_time'] != df['start_time'].shift(-1)]

print(df)


现在df如下所示:

  start_time          end_time             value start_time_         end_time_
0 2017-01-09 21:49:55 2017-01-09 21:55:41  150.0 2017-01-09 21:49:55                 NaT
1 2017-01-09 21:55:41 2017-01-09 21:58:46    4.0                 NaT 2017-01-09 21:58:46
2 2017-01-09 22:00:55 2017-01-09 23:13:00  144.0 2017-01-09 22:00:55 2017-01-09 23:13:00


然后填写NA值:

df['start_time_'].fillna(method='ffill',inplace=True)
df['end_time_'].fillna(method='bfill',inplace=True)


使用start_time_end_time_列替换start_timeend_time列。并删除start_time_end_time_列:

df['start_time'] = df['start_time_']
df['end_time'] = df['end_time_']
del df['start_time_']
del df['end_time_']


然后groupby和求和:

df = df.groupby(['start_time', 'end_time'], as_index=False).sum()

print(df)


结果如下:

           start_time            end_time  value
0 2017-01-09 21:49:55 2017-01-09 21:58:46  154.0
1 2017-01-09 22:00:55 2017-01-09 23:13:00  144.0

关于python - 如何将 Pandas DataFrame中的2行与连续的时间戳合并?,我们在Stack Overflow上找到一个类似的问题:https://stackoverflow.com/questions/42141875/

10-16 22:52