无条件匹配时使用默认值重新采样

Question

我通过重采样聚合数据，如下所示：

import quandl
import numpy as np

data = quandl.get("WIKI/KO", trim_start = "2000-12-12", trim_end = "2014-12-30")


data = data.ix[:40, ['Close']]

data['SIGNAL'] = np.random.randint(0,3, size=len(data))
data['SIGNAL'] = np.where((data['SIGNAL'] == 2),      -1, data['SIGNAL'] )
data['SIGNAL'] = np.where((data.index >= '2001-02-01'),      0, data['SIGNAL'] )

data['WIN'] = 10

print(data.to_string())


WinPerYear = data['WIN'].loc[(data['SIGNAL'] != 0)].resample('M').sum()
CntPerYear = data['WIN'].loc[(data['SIGNAL'] != 0)].resample('M').count()


print(WinPerYear.to_string())
print(CntPerYear.to_string())

正确地，我得到以下结果：

          Close  SIGNAL  WIN
Date                          
[...]
2001-01-17  57.94      -1   10
2001-01-18  57.13      -1   10
2001-01-19  55.81      -1   10
2001-01-22  55.69      -1   10
2001-01-23  56.88       0   10
2001-01-24  58.06      -1   10
2001-01-25  58.63      -1   10
2001-01-26  57.94      -1   10
2001-01-29  57.12       0   10
2001-01-30  57.91       0   10
2001-01-31  58.00       1   10
2001-02-01  57.44       0   10
2001-02-02  57.74       0   10
2001-02-05  59.20       0   10
2001-02-06  59.42       0   10
2001-02-07  60.00       0   10
2001-02-08  60.61       0   10
Date
2000-12-31     60
2001-01-31    160
Freq: M
Date
2000-12-31     6
2001-01-31    16
Freq: M

有没有一种简单的方法，即不改变所有的子集逻辑，也为所有不匹配的月份添加行？例如2001/02没有匹配项，因此我希望两个聚合都使用0，例如：

Date
2000-12-31     60
2001-01-31    160
2001-02-31    0
Freq: M
Date
2000-12-31     6
2001-01-31    16
2001-01-31    0
Freq: M

非常感谢并致以最良好的祝愿， e

Answer 1

我通过添加一个新的计算字段来解决这个问题

data['WIN_SIG']=0
data['WIN_SIG'][(data[Signal] != 0)] = data[Win]

在我聚合之后就被丢弃了

谢谢，干杯， E