Pandas基本操作-多级索引

Pandas基本操作-数据结构

多级索引创建和获取

关键就是是否存在一个可构建二维（多维）的index

1. 使用MultiIndex基于元组创建

创建一个Series数组，以元组作为index

import pandas as pd

index = [('California', 2000), ('California', 2010), 
    ('New York', 2000), ('New York', 2010), ('Texas', 2000), ('Texas', 2010)]
populations = [33871648, 37253956, 18976457, 19378102, 20851820, 25145561]
pop = pd.Series(populations, index=index)
print(pop)

结果如图

我们希望通过元组构建二级索引，使用pd.MultiIndex.from_tuples(index)

index1 = pd.MultiIndex.from_tuples(index)  # 基于元组创建
pop1 = pop.reindex(index1)
print(pop1)

是不是瞬间舒服很多，下面根据索引取值

print(pop1.loc['California'])  # 查询第一层索引
print("-------------------")
print(pop1.loc[:, 2010])  # 第二层索引
print(pop1.loc['California', 2010])  # 取一个元素

2. 通过二维索引数组创建

import numpy as np
import pandas as pd

index = [['a', 'a', 'b', 'b'], [1, 2, 1, 2]]
df = pd.DataFrame(np.random.rand(4, 2), index=index, columns=['data1', 'data2'])
print(df)

 或者使用pd.MultiIndex.from_arrays(二维数组)

index = pd.MultiIndex.from_arrays([['a', 'a', 'b', 'b'], [1, 2, 1, 2]])

取值，df[]从列开始，df.loc从索引（index）开始，.iloc基于索引位置开始的

print(df['data1'])         # 取'data1'列
print(df['data1']['a'])    # 取'data1'列索引'a'
print(df['data1'][:, 1])   # 取'data1'列索引1
print(df['data1']['a', 1]) # 取'data1'列索引'a'索引1

 结果

a  1    0.174784
   2    0.972091
b  1    0.565805
   2    0.020060
Name: data1, dtype: float64
1    0.174784
2    0.972091
Name: data1, dtype: float64
a    0.174784
b    0.565805
Name: data1, dtype: float64
0.174783518383268

获取索引为2的数据，使用切片索引

idx = pd.IndexSlice
print(df.loc[idx[:, 2] , :])

 结果

        data1     data2
a 2  0.865335  0.489147
b 2  0.350173  0.574875

 3. 笛卡尔积创建

pd.MultiIndex.from_product()

import numpy as np
import pandas as pd

index = pd.MultiIndex.from_product([[2013, 2014], [1, 2]], names=['year', 'visit'])  # 给索引起名字
columns = pd.MultiIndex.from_product([['Bob', 'Guido', 'Sue'], ['HR', 'Temp']])
# 模拟数据
data = np.round(np.random.randn(4, 6), 1)
data[:, ::2] *= 10
data += 37
# 创建一个包含多级列索引的DataFrame
health_data = pd.DataFrame(data, index=index, columns=columns)
print(health_data)

运行结果 

             Bob       Guido         Sue      
              HR  Temp    HR  Temp    HR  Temp
year visit                                    
2013 1      51.0  38.2  35.0  37.2  42.0  36.7
     2      38.0  37.4  59.0  35.4  44.0  36.1
2014 1      40.0  36.7  41.0  36.7  42.0  37.2
     2      47.0  35.7  31.0  37.4  38.0  37.3

“隔空取物”

idx = pd.IndexSlice
print(health_data.loc[:, idx[:, 'HR']])  # 列索引"HR"
print(health_data.loc[idx[:, 1], :])  # 索引1
print(health_data.loc[idx[:, 1], idx[:, 'HR']])  # 列索引"HR",索引1

运行结果

             Bob Guido   Sue
              HR    HR    HR
year visit                  
2013 1      26.0  39.0  37.0
     2      47.0  42.0  52.0
2014 1      48.0  33.0  29.0
     2      47.0  42.0  40.0
-----------------------------------------------
             Bob       Guido         Sue      
              HR  Temp    HR  Temp    HR  Temp
year visit                                    
2013 1      26.0  39.1  39.0  38.5  37.0  35.4
2014 1      48.0  36.2  33.0  38.3  29.0  35.8
-----------------------------------------------
             Bob Guido   Sue
              HR    HR    HR
year visit                  
2013 1      26.0  39.0  37.0
2014 1      48.0  33.0  29.0

4. stack() 和 unstack()

个人理解：

stack()：列索引只有一级
unstack()：行索引只有一级

运行上述代码 print(health_data.stack())

                  Bob  Guido   Sue
year visit                        
2013 1     HR    48.0   27.0  54.0
           Temp  38.3   36.1  38.1
     2     HR    40.0   44.0  47.0
           Temp  37.5   37.8  35.6
2014 1     HR    46.0   36.0  48.0
           Temp  37.4   38.7  37.2
     2     HR    34.0   35.0  44.0
           Temp  36.9   36.3  35.7

内容参考