Question

我有数据

data    id  url size    domain  subdomain
13/Jun/2016:06:27:26    30055   https://api.weather.com/v1/geocode/55.740002/37.610001/aggregate.json?apiKey=e45ff1b7c7bda231216c7ab7c33509b8&products=conditionsshort,fcstdaily10short,fcsthourly24short,nowlinks  3929    weather.com api.weather.com
13/Jun/2016:06:27:26    30055   https://api.weather.com/v1/geocode/54.720001/20.469999/aggregate.json?apiKey=e45ff1b7c7bda231216c7ab7c33509b8&products=conditionsshort,fcstdaily10short,fcsthourly24short,nowlinks  3845    weather.com api.weather.com
13/Jun/2016:06:27:27    3845    https://api.weather.com/v1/geocode/54.970001/73.370003/aggregate.json?apiKey=e45ff1b7c7bda231216c7ab7c33509b8&products=conditionsshort,fcstdaily10short,fcsthourly24short,nowlinks  30055   weather.com api.weather.com
13/Jun/2016:06:27:27    30055   https://api.weather.com/v1/geocode/59.919998/30.219999/aggregate.json?apiKey=e45ff1b7c7bda231216c7ab7c33509b8&products=conditionsshort,fcstdaily10short,fcsthourly24short,nowlinks  3914    weather.com api.weather.com
13/Jun/2016:06:27:28    30055   https://facebook.com    4005    facebook.com    facebook.com

我需要间隔5分钟对它进行分组。欲望输出

 data   id  url size    domain  subdomain
13/Jun/2016:06:27:26    30055   https://api.weather.com/v1/geocode/55.740002/37.610001/aggregate.json?apiKey=e45ff1b7c7bda231216c7ab7c33509b8&products=conditionsshort,fcstdaily10short,fcsthourly24short,nowlinks  3929    weather.com api.weather.com
13/Jun/2016:06:27:27    3845    https://api.weather.com/v1/geocode/54.970001/73.370003/aggregate.json?apiKey=e45ff1b7c7bda231216c7ab7c33509b8&products=conditionsshort,fcstdaily10short,fcsthourly24short,nowlinks  30055   weather.com api.weather.com
13/Jun/2016:06:27:28    30055   https://facebook.com    4005    facebook.com    facebook.com

我需要id, subdomain分组并建立间隔5min 我尝试使用

print df.groupby([df['data'],pd.TimeGrouper(freq='Min')])

以分钟为先分组，但返回TypeError: Only valid with DatetimeIndex, TimedeltaIndex or PeriodIndex, but got an instance of 'RangeIndex'

Answer 1

您需要使用data使用pd.to_datetime()设置解析format，并将结果用作index。然后.groupby()重新采样到5Min间隔：

df.index = pd.to_datetime(df.data, format='%d/%b/%Y:%H:%M:%S')
df.groupby(pd.TimeGrouper('5Min')).apply(lambda x: x.groupby(['id', 'subdomain']).first())

                                                           data  \
data                id    subdomain                               
2016-06-13 06:25:00 3845  api.weather.com  13/Jun/2016:06:27:27   
                    30055 api.weather.com  13/Jun/2016:06:27:26   
                          facebook.com     13/Jun/2016:06:27:28   

                                                                                         url  \
data                id    subdomain                                                            
2016-06-13 06:25:00 3845  api.weather.com  https://api.weather.com/v1/geocode/54.970001/7...   
                    30055 api.weather.com  https://api.weather.com/v1/geocode/55.740002/3...   
                          facebook.com                                  https://facebook.com   

                                            size        domain  
data                id    subdomain                             
2016-06-13 06:25:00 3845  api.weather.com  30055   weather.com  
                    30055 api.weather.com   3929   weather.com  
                          facebook.com      4005  facebook.com

Answer 2

请注意，要转换为datetime，您可以传递以下格式：

df['data'] = pd.to_datetime(df['data'], format="%d/%b/%Y:%H:%M:%S")

现在您可以使用groupby：

In [11]: df1 = df.set_index("data")

In [12]: df1.groupby(pd.TimeGrouper("5Min")).sum()
Out[12]:
                         id   size
data
2016-06-13 06:25:00  124065  45748

最好将其作为重新样本编写：

In [13]: df1.resample("5Min").sum()
Out[13]:
                         id   size
data
2016-06-13 06:25:00  124065  45748

Answer 3

首先需要set_index 检查df.index是日期时间索引。如果没有，那就是错误的原因

使用pandas按日期时间间隔分组

3 个答案: