python - Pandas HDFStore 从嵌套列中选择

Question

我有以下 DataFrame，它作为名为数据的 frame_table 存储在 HDFStore 对象中:

      shipmentid qty            
catid              1  2  3  4  5
0              0   0  0  0  0  0
1              1   0  0  0  2  0
2              2   2  0  0  0  0
3              3   0  4  0  0  0
0              0   0  0  0  0  0

我想执行 store.select('data','shipmentid==2')，但我收到“shipmentid”未定义的错误:

ValueError: The passed where expression: shipmentid==2
            contains an invalid variable reference
            all of the variable refrences must be a reference to
            an axis (e.g. 'index' or 'columns'), or a data_column
            The currently defined references are: columns,index

编写此 select 语句的正确方法是什么？

编辑:添加示例代码

import pandas as pd
from pandas import *
import random

def createFrame():
    data = {
             ('shipmentid',''):{1:1,2:2,3:3},
             ('qty',1):{1:5,2:5,3:5},
             ('qty',2):{1:6,2:6,3:6},
             ('qty',3):{1:7,2:7,3:7}
           }
    frame = pd.DataFrame(data)

    return frame

def createStore():
    store = pd.HDFStore('sample.h5',format='table')
    return store    

frame = createFrame()
print(frame)
print('\n')
print(frame.info())

store = createStore()
store.put('data',frame,format='t')
print('\n')
print(store)

results = store.select('data','shipmentid == 2')

store.close()

Answer 1

我敢打赌，您已经使用过类似的东西来创建您的商店，

In [207]:

data = pd.DataFrame(np.random.randn(8,2), columns=['shipmentid', 'qty'])
store = pd.HDFStore('borrar')
store.put('data', data, format='t')

如果您随后尝试执行 select 确实会收到您描述的错误，

In [208]:

store.select('data', 'shipmentid>0')

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-211-5d0c4082cdcf> in <module>()
----> 1 store.select('data', 'shipmentid>0')

...

ValueError: The passed where expression: shipmentid>0
            contains an invalid variable reference
            all of the variable refrences must be a reference to

相反，您可以这样创建它:

In [209]:

data = pd.DataFrame(np.random.randn(8,2), columns=['shipmentid', 'qty'])
data.to_hdf('borrar2', 'data', append=True, mode='w', data_columns=['shipmentid', 'qty'])
In [210]:

pd.read_hdf('borrar2', 'data', where='shipmentid>0')
Out[210]:
shipmentid  qty
1   0.778225    -1.008529
5   0.264075    -0.651268
7   0.908880    0.153306

(老实说，我不知道为什么它以一种方式工作而另一种不工作，我的猜测是在第一种方式中你不能指定数据列。但这是其中一个可以驱使你疯了……)

编辑:更新发布的代码后，数据框有一个 MultiIndex。类似的更新代码类似于:

In [273]:

import pandas as pd
from pandas import *
import random

def createFrame():
    data = {
             ('shipmentid',''):{1:1,2:2,3:3},
             ('qty',1):{1:5,2:5,3:5},
             ('qty',2):{1:6,2:6,3:6},
             ('qty',3):{1:7,2:7,3:7}
           }
    frame = pd.DataFrame(data)

    return frame 

frame = createFrame()
print(frame)
print('\n')
print(frame.info())

frame.to_hdf('sample.h5', 'data', append=True, mode='w', data_columns=['shipmentid'], format='table')
pd.read_hdf('sample.h5','data', 'shipmentid == 2')

但是我得到了一个错误(我猜你也得到了同样的结果):

  qty       shipmentid
    1  2  3           
1   5  6  7          1
2   5  6  7          2
3   5  6  7          3


<class 'pandas.core.frame.DataFrame'>
Int64Index: 3 entries, 1 to 3
Data columns (total 4 columns):
(qty, 1)          3 non-null int64
(qty, 2)          3 non-null int64
(qty, 3)          3 non-null int64
(shipmentid, )    3 non-null int64
dtypes: int64(4)
memory usage: 120.0 bytes
None
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-273-e10e811fc7c0> in <module>()
     23 print(frame.info())
     24 
---> 25 frame.to_hdf('sample.h5', 'data', append=True, mode='w', data_columns=['shipmentid'], format='table')
     26 pd.read_hdf('sample.h5','data', 'shipmentid == 2')
.....
stack trace
.....
ValueError: cannot use a multi-index on axis [1] with data_columns ['shipmentid']

我已经浏览了一下，但我无法为此提供解决方案。我的印象是通过查看 code in github是选项 data_columns 不能与 MultiIndex 结合使用。我能想到的唯一解决方案是像在您的代码中一样写入 HDFStore ，然后读取完整的数据帧，不带任何条件，然后进行搜索。即:

new_frame = store.get('data')
print new_frame[new_frame['shipmentid'] == 2]



<class 'pandas.io.pytables.HDFStore'>
File path: sample.h5
/data            frame_table  (typ->appendable,nrows->3,ncols->4,indexers->[index])
  qty       shipmentid
    1  2  3           
2   5  6  7          2