[Data analysis visualization] Multi-level index

import numpy as np
import pandas as pd
from pandas import Series, DataFrame

s1 = Series(np.random.rand(6))
s1

0    0.710042
1    0.901424
2    0.050802
3    0.870486
4    0.919496
5    0.483373
dtype: float64

# 创建多级index(相当于Series的list，里面被划分为1，2两个Series)
s1 = Series(np.random.rand(6), index=[[1,1,1,2,2,2],['a','b','c','a','b','c']])
s1

1  a    0.005413
   b    0.668101
   c    0.540828
2  a    0.922140
   b    0.046360
   c    0.207378
dtype: float64

s1[1]

a    0.005413
b    0.668101
c    0.540828
dtype: float64

# 多级index的其中一级 Series类型
type(s1[1])

pandas.core.series.Series

# 取值
s1[1]['a']

0.005413335166173483

# 可以切片取值(不同级的同一个key)
s1[:,'a']

1    0.005413
2    0.922140
dtype: float64

# 切片的多级index也是Series类型
type(s1[:,'a'])

pandas.core.series.Series

Multilevel index conversion to DataFrame

# 转换DataFrame
df1 = s1.unstack()
df1

	a	b	c
1	0.005413	0.668101	0.540828
2	0.922140	0.046360	0.207378

# 由两个Series组成DataFrame
df2 = DataFrame([s1[1],s1[1]])
df2

	a	b	c
0	0.005413	0.668101	0.540828
1	0.005413	0.668101	0.540828

Convert DataFrame into multi-level index

# 转化（但分级不对）
s2 = df1.unstack()
s2

a  1    0.005413
   2    0.922140
b  1    0.668101
   2    0.046360
c  1    0.540828
   2    0.207378
dtype: float64

# 转化（解决分级不对，转置T）
s2 = df1.T.unstack()
s2

1  a    0.005413
   b    0.668101
   c    0.540828
2  a    0.922140
   b    0.046360
   c    0.207378
dtype: float64

Create a multilevel index DataFrame

df2 = DataFrame(np.arange(16).reshape(4,4))
df2

	0	1	2	3
0	0	1	2	3
1	4	5	6	7
2	8	9	10	11
3	12	13	14	15

# 通过index,columns分级
df2 = DataFrame(np.arange(16).reshape(4,4), index=[['a','a','b','b'],[1,2,1,2]],columns=[['BJ','BJ','SH','SH'],[4,6,4,6]])
df2

		BJ		SH
		4	6	4	6
a	1	0	1	2	3
	2	4	5	6	7
b	1	8	9	10	11
	2	12	13	14	15

# 取数据默认是 列索引 输出
df2['BJ']

		4	6
a	1	0	1
	2	4	5
b	1	8	9
	2	12	13

type(df2['BJ'])

pandas.core.frame.DataFrame

df2['BJ'][4]

a  1     0
   2     4
b  1     8
   2    12
Name: 4, dtype: int64