我遇到了 pandas HDFStore 方法的问题，我无法以使用 h5py.File 方法检索的方式访问数据。这是代码片段:
In [1]: import pandas as pd
In [2]: import numpy as np
In [3]: import h5py as h5
In [4]: hdf = pd.HDFStore("tmp.h5")
In [5]: hdf.put('tables/t1', pd.DataFrame(np.random.rand(20,5)))
In [6]: hdf.put('t2', pd.DataFrame(np.random.rand(10,5)))
In [7]:
In [7]: hdf.close()
In [8]:
In [8]: ############ Read using pd.HDFStore ############
In [9]:
In [9]: data = pd.HDFStore ("tmp.h5")
In [10]: data["tables/t1"]
Out[10]:
0 1 2 3 4
0 0.384926 0.712066 0.022438 0.686217 0.942678
1 0.079548 0.466799 0.575394 0.276646 0.514414
2 0.672582 0.828567 0.801799 0.296046 0.124042
3 0.568058 0.931348 0.225348 0.547913 0.736184
4 0.496768 0.419699 0.724118 0.313427 0.353825
5 0.771868 0.963346 0.523821 0.793295 0.052085
6 0.358478 0.845149 0.334389 0.674448 0.239096
7 0.454559 0.604438 0.183654 0.027641 0.186922
8 0.776586 0.155783 0.253801 0.123986 0.560601
9 0.201239 0.932080 0.040997 0.119049 0.154076
10 0.753566 0.770133 0.123285 0.112419 0.353622
11 0.040959 0.384800 0.806119 0.247106 0.013442
12 0.739205 0.100547 0.855418 0.774874 0.710557
13 0.865856 0.565094 0.815860 0.816869 0.834415
14 0.251312 0.624995 0.976317 0.854855 0.744861
15 0.179678 0.435902 0.602303 0.118516 0.386935
16 0.452009 0.973729 0.067736 0.097811 0.292619
17 0.285994 0.569845 0.584602 0.001671 0.422877
18 0.727996 0.291086 0.736912 0.960595 0.132891
19 0.356397 0.747693 0.458485 0.100849 0.072220
In [11]: ## Success
In [12]: data ["tables"]["t1"]
---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
<ipython-input-12-c7599d16a7b6> in <module>()
----> 1 data ["tables"]["t1"]
/usr/conda/lib/python2.7/site-packages/pandas/io/pytables.py in __getitem__(self, key)
415
416 def __getitem__(self, key):
--> 417 return self.get(key)
418
419 def __setitem__(self, key, value):
/usr/conda/lib/python2.7/site-packages/pandas/io/pytables.py in get(self, key)
632 if group is None:
633 raise KeyError('No object named %s in the file' % key)
--> 634 return self._read_group(group)
635
636 def select(self, key, where=None, start=None, stop=None, columns=None,
/usr/conda/lib/python2.7/site-packages/pandas/io/pytables.py in _read_group(self, group, **kwargs)
1268
1269 def _read_group(self, group, **kwargs):
-> 1270 s = self._create_storer(group)
1271 s.infer_axes()
1272 return s.read(**kwargs)
/usr/conda/lib/python2.7/site-packages/pandas/io/pytables.py in _create_storer(self, group, format, value, append, **kwargs)
1151 else:
1152 raise TypeError(
-> 1153 "cannot create a storer if the object is not existing "
1154 "nor a value are passed")
1155 else:
TypeError: cannot create a storer if the object is not existing nor a value are passed
In [13]:
In [13]: data.close()
In [14]:
In [14]: ########### Read using h5py.File ##############
In [15]:
In [15]: data = h5.File("tmp.h5","r")
In [16]:
In [16]: data["tables"]
Out[16]: <HDF5 group "/tables" (1 members)>
In [17]:
In [17]: data["tables"]["t1"]
Out[17]: <HDF5 group "/tables/t1" (4 members)>
In [18]:
In [18]: data['tables']['t1'].keys ()
Out[18]: [u'axis0', u'axis1', u'block0_items', u'block0_values']
In [19]: [u'axis0', u'axis1', u'block0_items', u'block0_values']
Out[19]: [u'axis0', u'axis1', u'block0_items', u'block0_values']
In [20]:
In [20]: data['tables']['t1']['block0_values'].value
Out[20]:
array([[ 0.38492571, 0.71206567, 0.02243773, 0.68621713, 0.9426783 ],
[ 0.07954806, 0.4667994 , 0.57539433, 0.27664603, 0.51441446],
[ 0.67258161, 0.82856681, 0.80179916, 0.29604625, 0.12404214],
[ 0.56805845, 0.93134797, 0.22534757, 0.54791294, 0.73618366],
[ 0.49676792, 0.41969943, 0.72411835, 0.31342698, 0.35382463],
[ 0.77186804, 0.96334586, 0.52382094, 0.7932945 , 0.05208528],
[ 0.3584784 , 0.84514863, 0.33438851, 0.6744483 , 0.23909552],
[ 0.45455901, 0.6044383 , 0.18365449, 0.02764097, 0.18692162],
[ 0.77658631, 0.15578276, 0.25380109, 0.12398617, 0.56060138],
[ 0.20123928, 0.93207974, 0.04099724, 0.11904895, 0.15407568],
[ 0.75356644, 0.77013349, 0.12328475, 0.11241904, 0.35362213],
[ 0.04095888, 0.38480023, 0.80611853, 0.24710571, 0.01344193],
[ 0.73920528, 0.1005474 , 0.85541761, 0.7748739 , 0.71055697],
[ 0.86585587, 0.5650938 , 0.81586031, 0.81686915, 0.83441517],
[ 0.25131205, 0.62499501, 0.97631707, 0.85485518, 0.74486096],
[ 0.17967805, 0.43590236, 0.60230302, 0.11851596, 0.38693535],
[ 0.4520091 , 0.97372923, 0.0677363 , 0.09781059, 0.29261929],
[ 0.28599448, 0.56984462, 0.5846021 , 0.00167063, 0.42287738],
[ 0.72799625, 0.29108631, 0.7369122 , 0.96059508, 0.13289119],
[ 0.35639696, 0.7476934 , 0.45848456, 0.10084881, 0.07221995]])
In [21]:
In [21]: ######################## End ###############
In [22]:
In [22]:
我想使用data['tables']['t1']方式访问数据。我因为这个问题而陷入困境。我观察到 pandas 将 hd5 中的每个数据帧作为组插入。我想将其作为数据集插入，以便我可以轻松访问数据。
最佳答案
根据documentation对于HDFStore。
Warning: Hierarchical keys cannot be retrieved as dotted (attribute) access as described above for items stored under the root node. Instead, use explicit string based keys.
所以这只是模块加载 hdfs 的方式不同。
关于python - 将 pandas 数据框作为数据集插入 HDFStore 中，我们在Stack Overflow上找到一个类似的问题： https://stackoverflow.com/questions/42739415/