Pandas：windows 数组的滚动平均值

Question

类似于，我可以计算多个rolling means

d1 = df.set_index('DateTime').sort_index()
ma_1h = d1.groupby('Event').rolling('H').mean()
ma_2h = d1.groupby('Event').rolling('2H').mean()

但是如果我想对数组列表执行此操作，我该如何高效地执行此操作？

window_array = ['H','3H','6H','9H'] # etc

并且我的滚动方式被包含回我的原始数据框中

Answer 1

我相信您需要转换偏移量并通过列表理解在循环中创建新的 DataFrames，最后 concat:

from pandas.tseries.frequencies import to_offset

df1 = pd.concat([d1.groupby('Event').rolling(to_offset(x)).mean() for x in window_array], 
                axis=1, 
                keys=window_array)

样本：

rng = pd.date_range('2017-04-03', periods=10, freq='38T')
df = pd.DataFrame({'DateTime': rng, 'a': range(10), 'Event':[4] * 3 + [3] * 3 + [1] * 4})  
print (df)


from pandas.tseries.frequencies import to_offset
window_array = ['H','3H','6H','9H']


d1 = df.set_index('DateTime').sort_index()
a = pd.concat([d1.groupby('Event')['a'].rolling(to_offset(x)).mean() for x in window_array], 
              axis=1, 
              keys=window_array)
print (a)
                             H   3H   6H   9H
Event DateTime                               
1     2017-04-03 03:48:00  6.0  6.0  6.0  6.0
      2017-04-03 04:26:00  6.5  6.5  6.5  6.5
      2017-04-03 05:04:00  7.5  7.0  7.0  7.0
      2017-04-03 05:42:00  8.5  7.5  7.5  7.5
3     2017-04-03 01:54:00  3.0  3.0  3.0  3.0
      2017-04-03 02:32:00  3.5  3.5  3.5  3.5
      2017-04-03 03:10:00  4.5  4.0  4.0  4.0
4     2017-04-03 00:00:00  0.0  0.0  0.0  0.0
      2017-04-03 00:38:00  0.5  0.5  0.5  0.5
      2017-04-03 01:16:00  1.5  1.0  1.0  1.0

Answer 2

window_array = ['H','3H','6H','9H'] # etc
for window in window_array:
    d1[window] = d1.groupby('Event').rolling(window).mean()

Pandas：windows 数组的滚动平均值

Pandas: Rolling mean over array of windows

python

time-series

moving-average

pandas