Python 3.x 如何将时间序列数据捆绑到熊猫的24小时间隔中？

我有一个csv文件，其中包含一些数据

,date,location,device,provider,cpu,mem,load,drops,id,latency,gw_latency,upload,download,sap_drops,sap_latency,alert_id
389,2018-02-13 09:20:17.572685+00:00,ASA,10.11.100.1,BOM,4.0,23.0,0.25,0.0,,,,,,,,
390,2018-02-13 09:20:21.836284+00:00,ASA,10.11.100.1,COD,4.0,23.0,2.08,0.0,,,,,,,,
391,2018-02-13 09:30:59.401178+00:00,ASA,10.11.100.1,COD,5.0,23.0,8.0,0.0,,,,,,,,
392,2018-02-13 09:31:03.667730+00:00,ASA,10.11.100.1,COD,5.0,23.0,3.5,0.0,,,,,,,,
393,2018-02-13 09:41:14.666626+00:00,ASA,10.11.100.1,BOM,4.0,23.0,0.5,0.0,,,,,,,,
394,2018-02-13 09:41:18.935061+00:00,ASA,10.11.100.1,DAE,4.0,23.0,3.0,0.0,,,,,,,,
395,2018-02-13 09:50:17.491014+00:00,ASA,10.11.100.1,DAE,5.0,23.0,8.25,0.0,,,,,,,,
396,2018-02-13 09:50:21.751805+00:00,BBB,10.11.100.1,BOM,5.0,23.0,2.75,0.0,,,,,,,,
397,2018-02-13 10:00:18.387647+00:00,BBB,10.11.100.1,CXU,5.0,23.0,2.0,0.0,,,,,,,,
398,2018-02-13 10:00:22.847626+00:00,ASA,10.11.100.1,BOM,5.0,23.0,3.17,0.0,,,,,,,,
399,2018-02-13 10:10:17.521642+00:00,BBB,10.11.100.1,DAE,5.0,23.0,1.0,0.0,,,,,,,,
400,2018-02-13 10:10:21.786720+00:00,BBB,10.11.100.1,DAE,5.0,23.0,2.42,0.0,,,,,,,,
401,2018-02-13 10:14:38.085999+00:00,BBB,10.11.100.1,CXU,4.0,23.0,0.25,0.0,,,,,,,,
..................................................................................
..................................................................................

正如您所看到的，日期

2018-02-13

在几个时间间隔内有很多条目。我想将这些条目放入

24

小时间隔中，每个小时将包含一个值（平均值）。这就是我所做的

df_next = df.loc['2018-04-13'].resample('H')["cpu"].mean().fillna(0)

但是，对于日期

2018-04-13

，收集的数据最多只有

10:00

小时（最后一次输入是在

10:14:38

），所以它只给出了最多的时间。对于其他一些日期，如果数据是从

9:00

开始收集的，那么我只得到从

9:00

开始的每小时间隔

无论在什么时候收集数据，我如何从日期的

00:00

开始获得完整的

24

小时间隔？因此，基本上，它将为未收集数据的小时分配

0

，为收集数据的小时分配

mean

所以基本上我想要这样的东西

381,2018-02-13 00:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 
382,2018-02-13 01:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 
383,2018-02-13 02:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
384,2018-02-13 03:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 
385,2018-02-13 04:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 
386,2018-02-13 05:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
387,2018-02-13 06:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 
388,2018-02-13 07:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
388,2018-02-13 08:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 

389,2018-02-13 09:20:17.572685+00:00,ASA,10.11.100.1,BOM,4.0,23.0,0.25,0.0,,,,,,,,
390,2018-02-13 09:20:21.836284+00:00,ASA,10.11.100.1,COD,4.0,23.0,2.08,0.0,,,,,,,,
391,2018-02-13 09:30:59.401178+00:00,ASA,10.11.100.1,COD,5.0,23.0,8.0,0.0,,,,,,,,
392,2018-02-13 09:31:03.667730+00:00,ASA,10.11.100.1,COD,5.0,23.0,3.5,0.0,,,,,,,,
393,2018-02-13 09:41:14.666626+00:00,ASA,10.11.100.1,BOM,4.0,23.0,0.5,0.0,,,,,,,,
394,2018-02-13 09:41:18.935061+00:00,ASA,10.11.100.1,DAE,4.0,23.0,3.0,0.0,,,,,,,,
395,2018-02-13 09:50:17.491014+00:00,ASA,10.11.100.1,DAE,5.0,23.0,8.25,0.0,,,,,,,,
396,2018-02-13 09:50:21.751805+00:00,BBB,10.11.100.1,BOM,5.0,23.0,2.75,0.0,,,,,,,,
397,2018-02-13 10:00:18.387647+00:00,BBB,10.11.100.1,CXU,5.0,23.0,2.0,0.0,,,,,,,,
398,2018-02-13 10:00:22.847626+00:00,ASA,10.11.100.1,BOM,5.0,23.0,3.17,0.0,,,,,,,,
399,2018-02-13 10:10:17.521642+00:00,BBB,10.11.100.1,DAE,5.0,23.0,1.0,0.0,,,,,,,,
400,2018-02-13 10:10:21.786720+00:00,BBB,10.11.100.1,DAE,5.0,23.0,2.42,0.0,,,,,,,,
401,2018-02-13 10:14:38.085999+00:00,BBB,10.11.100.1,CXU,4.0,23.0,0.25,0.0,,,,,,,,

402,2018-02-13 11:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 
403,2018-02-13 12:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 
404,2018-02-13 13:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
405,2018-02-13 14:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 
406,2018-02-13 15:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 
407,2018-02-13 16:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
408,2018-02-13 17:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 
409,2018-02-13 18:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 
410,2018-02-13 19:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 
411,2018-02-13 20:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 
412,2018-02-13 21:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 
413,2018-02-13 22:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 
414,2018-02-13 23:00:00.000000+00:00,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 

如您所见，它会填满剩余的时间，并将其值保持为

0

使用：

#get Series only for hourly data
#remove non exist hours by dropna 
a = df.resample('H')["cpu"].mean().dropna()
#create all posible hours by first min and max value floor to 0 and 23 hour
rng = pd.date_range(a.index.min().floor('d'), 
                    a.index.max().floor('d') + pd.Timedelta(23, unit='h'), freq='H')
#get all missing index values - missing hours
diff_idx = rng.difference(a.index)

#join new DataFrame with missing values to original, last sorting for correct ordering
df = pd.concat([df, pd.DataFrame(index=diff_idx, columns=df.columns)]).sort_index()

---

嘿，我想要一个特定日期的

24

小时间隔。在这种情况下，

2018-04-13

。而对于另一个数据，我不需要？基本上，我会得到一个日期，然后我必须给出该日期24小时间隔的

cpu

值。因此，特定日期将有24个值。@SouvikRay-可以删除

NaN

s行-不存在的间隔？使用

s.fillna（0，inplace=True）

而不是

s.dripna（）