一、创建多层次索引

（1）隐式构造

• 最常见的方法是给DataFrame构造函数的index参数传递两个或更多的数组

# 导入pandas
import numpy as np
import pandas as pd

data = np.random.randint(0,100,size=(6,6))

# 行索引
index = [
    ["1班","1班","1班","2班","2班","2班"],
    ["张三","李四","王五","鲁班","张三丰","张无忌"]
]

# 列索引
columns = [
    ["期中","期中","期中","期末","期末","期末"],
    ["语文","数学","英语","语文","数学","英语"]
]

df = pd.DataFrame(data=data,index=index,columns=columns)
df

• Series也可以创建多层索引

data = np.random.randint(0,100,size=6)

index = [
    ["1班","1班","1班","2班","2班","2班"],
    ["张三","李四","王五","鲁班","张三丰","张无忌"]
]

s = pd.Series(data=data,index=index)
s

（2）显式构造pd.MultiIndex

• 使用数组

data = np.random.randint(0,100,size=(6,6))

# 行索引
index = pd.MultiIndex.from_arrays([
    ["1班","1班","1班","2班","2班","2班"],
    ["张三","李四","王五","鲁班","张三丰","张无忌"]
])

# 列索引
columns = [
    ["期中","期中","期中","期末","期末","期末"],
    ["语文","数学","英语","语文","数学","英语"]
]

df = pd.DataFrame(data=data,index=index,columns=columns)
df

• 使用tuple

data = np.random.randint(0,100,size=(6,6))

# 行索引
index = pd.MultiIndex.from_tuples(
    (
        ("1班","张三"),("1班","李四"),("1班","王五"),
        ("2班","鲁班"),("2班","张三丰"),("2班","张无忌")
    )
)

# 列索引
columns = [
    ["期中","期中","期中","期末","期末","期末"],
    ["语文","数学","英语","语文","数学","英语"]
]

df = pd.DataFrame(data=data,index=index,columns=columns)
df

• 使用product

笛卡尔积：{a,b}{c,d} ==> {a,c},{a,d},{b,c},{b,d}

data = np.random.randint(0,100,size=(6,6))

# 行索引
index = pd.MultiIndex.from_product([
    ["1班","2班"],
    ["张三","李四","王五"]
])

# 列索引
columns = [
    ["期中","期中","期中","期末","期末","期末"],
    ["语文","数学","英语","语文","数学","英语"]
]

df = pd.DataFrame(data=data,index=index,columns=columns)
df

二、多层列索引

除了行索引index，列索引columns也能用同样的方法创建多层索引

• 使用数组

data = np.random.randint(0,100,size=(6,6))

# 行索引
index = pd.MultiIndex.from_arrays([
    ["1班","1班","1班","2班","2班","2班"],
    ["张三","李四","王五","鲁班","张三丰","张无忌"]
])

# 列索引
columns = pd.MultiIndex.from_arrays([
    ["期中","期中","期中","期末","期末","期末"],
    ["语文","数学","英语","语文","数学","英语"]
])

df = pd.DataFrame(data=data,index=index,columns=columns)
df

• 使用tuple

data = np.random.randint(0,100,size=(6,6))

# 行索引
index = pd.MultiIndex.from_tuples(
    (
        ("1班","张三"),("1班","李四"),("1班","王五"),
        ("2班","鲁班"),("2班","张三丰"),("2班","张无忌")
    )
)

# 列索引
columns = pd.MultiIndex.from_tuples(
    (
        ("期中","语文"),("期中","数学"),("期中","英语"),
        ("期末","语文"),("期末","数学"),("期末","英语")
    )
)

df = pd.DataFrame(data=data,index=index,columns=columns)
df

• 使用product

data = np.random.randint(0,100,size=(6,6))

# 行索引
index = pd.MultiIndex.from_product([
    ["1班","2班"],
    ["张三","李四","王五"]
])

# 列索引
columns = pd.MultiIndex.from_product([
    ["期中","期末"],
    ["语文","数学","英语"]
])

df = pd.DataFrame(data=data,index=index,columns=columns)
df