Python 根据索引选择行

我试着用

data = data.loc['bids:']

获取与索引对应的所有行

文本文件中的示例数据：

 {"offset":"14726172634","bids":[["871094.22000","0.00200000","0.00200000","0","1081537351","29194","5","14726172633","1"],["871076.11000","0.00808000","0.00808000","0","1081537130","623964","5","14726172043","1"],["871073.96500","0.00100000","0.00100000","0","1081537185","29194","5","14726172231","1"]]],
"asks":[["875644.72000","0.00200000","0.00200000","0","1081606189","29194","5","14726356256","1"],["875669.77637","0.01000000","0.01000000","0","1081606227","29194","5","14726356379","1"],["875678.92000","0.00600000","0.00600000","0","1081606263","29194","5","14726356488","1"],["875731.74364","0.03000000","0.03000000","0","1081606233","29194","5","14726356393","1"],

代码示例：

 data = pd.read_csv('20190523-012523_product_5_snapshot_14726172634_14728561053.txt', lineterminator= str(']'), low_memory= False, error_bad_lines=False, header= None)#, names= ['a','d','f','r','y','h','n','m','k'])

 new = data[1].str.split("[", n = 1, expand = True)
 data[1]= new[0]
 data[10]= new[1]
 data.drop(data.index[-1], inplace=True)
 data[10]= new[1].str.strip('[').str.strip('"')

 data = data.set_index([1,2])
 data = data.loc[:,[10]]
 data = data.loc['bids:']

数据样本：

bids:  0.002000   871094.22000
       0.008080   871076.11000
       0.001000   871073.96500
bids:  0.005000   871042.87000
       0.005000   871038.55000
       0.001000   871032.90156

代码输出：

bids:  0.002000   871094.22000
bids:  0.005000   871042.87000

请问这6排怎么走？其目的是在其他索引标签之间进行筛选

索引输出为：

Index(['bids:', '', '', '', '', '', '', '', '', '',
       ...
       'asks:', '', '', '', '', '', '', '', '', '',
       ...
       'bids:', '', '', '', '', '', '', '', '', '',
       ...'],
      dtype='object', name=1, length=505148)

我相信你需要：

print (data)
                           10
1    2                       
bids 0.00200000  871094.22000
     0.00808000  871076.11000
     0.00100000  871073.96500
asks 0.00200000  875644.72000
     0.01000000  875669.77637
     0.00600000  875678.92000
     0.03000000  875731.74364
bids 0.00200000  871094.22000
     0.00808000  871076.11000
     0.00100000  871073.96500

print (data.index)
MultiIndex(levels=[['asks', 'bids'], 
                   ['0.00100000', '0.00200000', '0.00600000', 
                    '0.00808000', '0.01000000', '0.03000000']],
           codes=[[1, 1, 1, 0, 0, 0, 0, 1, 1, 1], [1, 3, 0, 1, 4, 2, 5, 1, 3, 0]],
           names=[1, 2])

按第一级

多索引

求解第一个值，具有重复值：

s = data.index.get_level_values(0).to_series()
mask = s.ne(s.shift())
print (mask)
1
bids     True
bids    False
bids    False
asks     True
asks    False
asks    False
asks    False
bids     True
bids    False
bids    False
Name: 1, dtype: bool

df = data[mask.values]
print (df)
                           10
1    2                       
bids 0.00200000  871094.22000
asks 0.00200000  875644.72000
bids 0.00200000  871094.22000

df = df.xs('bids', drop_level=False)
print (df)
                           10
1    2                       
bids 0.00200000  871094.22000
     0.00200000  871094.22000

如果没有

多索引

：

print (data)
              2             10
1                             
bids  0.00200000  871094.22000
bids  0.00808000  871076.11000
bids  0.00100000  871073.96500
asks  0.00200000  875644.72000
asks  0.01000000  875669.77637
asks  0.00600000  875678.92000
asks  0.03000000  875731.74364
bids  0.00200000  871094.22000
bids  0.00808000  871076.11000
bids  0.00100000  871073.96500

---

---

@ALollz只是添加了一个假设，当行索引为“”时，Pandas认为没有索引。不，它认为索引是空字符串。您可能想要创建一个多索引，其中第一个级别是['bids'，'asks'，]，第二个级别是第一个小数字。如果您可以包含创建数据帧的代码，或者至少是您正在显示的示例版本，这将有所帮助。例如，类似于

df=pd.DataFrame（{'some_column'：[1.0，2.0….

。这样，我们就知道你的数据框的结构是什么样的，我们可以复制粘贴代码来开发和测试我们的答案。现在，它没有以标准格式呈现，告诉我们你的数据到底发生了什么。谢谢！@Mike原始数据来自我解析的文本文件。你还记得吗你觉得应该贴出来吗？这会让这篇文章很难读。谢谢

print (data.index)
Index(['bids', 'bids', 'bids', 'asks', 'asks', 'asks', 'asks', 'bids', 'bids',
       'bids'],
      dtype='object', name=1)

s = data.index.to_series()
mask = s.ne(s.shift())
print (mask)
1
bids     True
bids    False
bids    False
asks     True
asks    False
asks    False
asks    False
bids     True
bids    False
bids    False
Name: 1, dtype: bool

df = data[mask.values].loc['bids']
print (df)
              2             10
1                             
bids  0.00200000  871094.22000
bids  0.00200000  871094.22000