Question

我想每天/每周/每月重新采样DataFrame。我很困惑，并且不知道它应该如何。如何使用条件重新采样并对新创建的行求和。

df = pd.DataFrame({ 
    'date': ['2014-08-4 19:00:00', '2014-08-5 10:09:00', '2014-08-4 21:04:00','2014-08-5 22:07:00', '2014-08-5 22:09:00', '2014-08-5 22:09:00', '2014-08-4 22:09:00', '2014-08-5 22:09:00', '2014-08-4 22:09:00', '2014-08-5 22:09:00', '2014-08-4 22:09:00', '2014-08-4 22:09:00', '2014-08-5 22:09:00', '2014-08-4 22:09:00', '2014-08-4 22:09:00', '2014-08-5 22:09:00', '2014-08-4 22:09:00', '2014-08-5 22:09:00', '2014-08-4 22:09:00', '2014-08-5 22:09:00'], 
    'id'  :[4,5,7,8,2,3,5,2,1,1,4,4,2,4,5,1,3,9,7,9],
    'qty' :[9,5,7,8,3,3,5,2,1,1,4,4,2,4,5,1,3,5,7,9], 
    'type' :[1,0,1,0,1,1,0,0,1,1,0,0,0,1,1,1,0,0,1,0] 
})
inward = df['type'] == 0
outward = df['type'] == 1

df1 = df.join(df[inward].groupby(['id'])['qty'].sum(), on='id', rsuffix='_inward')
df2 = df.join(df[outward].groupby(['id'])['qty'].sum(), on='id', rsuffix='_outward')

df1['qty_outward'] = df2['qty_outward']

我试图以下列格式获取数据

    date                 id qty_inward  qty_outward
0   2014-08-04 19:00:00  4  8           13
1   2014-08-05 10:09:00  5  5           0
2   2014-08-04 21:04:00  7  0           14
3   2014-08-05 22:07:00  8  8           0
4   2014-08-05 22:09:00  2  4           3
5   2014-08-05 22:09:00  3  0           3
8   2014-08-04 22:09:00  1  0           1
9   2014-08-05 22:09:00  1  0           2
14  2014-08-04 22:09:00  5  5           5
16  2014-08-04 22:09:00  3  3           0
17  2014-08-05 22:09:00  9  14          0

这些我每周/每日/每月创建开仓和平仓股票。如果您有任何建议，我的方法可能是错误的。

Answer 1

我认为您可以resample使用groupby - 这是0.18.1中的新功能。最后通过unstack第二级重塑到列：

print (df.groupby(['id', 'type'])
         .resample('D')['qty']
         .sum()
         .unstack(1, fill_value=0)
         .reset_index(level=0))

type        id  qt_inward  qt_outward
date                                 
2014-08-04   1          0           1
2014-08-05   1          0           2
2014-08-05   2          4           3
2014-08-04   3          3           0
2014-08-05   3          0           3
2014-08-04   4          8          13
2014-08-04   5          5           5
2014-08-05   5          5           0
2014-08-04   7          0          14
2014-08-05   8          8           0
2014-08-05   9         14           0

按日期重新采样并根据条件添加新的总和列

1 个答案: