python 数据分析实现长宽格式的转换

# encoding=utf-8

import numpy as np
import pandas as pd

# 长宽格式的转换
# 1
data = pd.read_csv(\'d:data/macrodata.csv\')
print \'data:=\\n\', data
print \'data.to_records():=\\n\', data.to_records()
print \'data.year:=\\n\', data.year
print \'data.quarter:=\\n\', data.quarter

periods = pd.PeriodIndex(year=data.year, quarter=data.quarter, name=\'date\')
print \'periods:=\\n\', periods
data = pd.DataFrame(data.to_records(),
          columns=pd.Index([\'realgdp\', \'infl\', \'unemp\'], name=\'item\'),
          index=periods.to_timestamp(\'D\', \'end\'))
print \'data:=\\n\', data
ldata = data.stack().reset_index().rename(columns={0: \'value\'})
# print \'ldata:=\\n\', ldata
print \'ldata.get(\\\'realgdp\\\'):=\\n\', ldata.get(\'realgdp\')
print \'ldata.get(\\\'unemp\\\'):=\\n\', ldata.get(\'unemp\')
wdata = ldata.pivot(\'date\', \'item\', \'value\')
print \'ldata:=\\n\', ldata
print \'wdata:=\\n\', wdata

# 2
print \'ldata[:10]:=\\n\', ldata[:10]
pivoted = ldata.pivot(\'date\', \'item\', \'value\')
print \'pivoted:=\\n\', pivoted
print \'pivoted.head():=\\n\', pivoted.head()
print \'ldata:=\\n\', ldata
ldata[\'value2\'] = np.random.randn(len(ldata))
print \'ldata[\\\'value2\\\']:=\\n\', ldata[\'value2\']
print \'ldata[:10]:=\\n\', ldata[:10]
pivoted = ldata.pivot(\'date\', \'item\')
print \'pivoted:=\\n\', pivoted
print pivoted[:5]
print \'pivoted[\\\'value\\\'][:5]:=\\n\', pivoted[\'value\'][:5]
print \'ldata:=\\n\', ldata
unstacked = ldata.set_index([\'date\', \'item\']).unstack(\'item\')
print \'unstacked:=\\n\', unstacked

print \'test\'

import pandas as pd
import numpy as np
 
 
# 常用的表格形式的数据结构
df = pd.DataFrame(np.arange(6).reshape((2,3)), index=[\'id1\',\'id2\'], columns=[\'attr1\',\'attr2\',\'attr3\'])
print(df)
out:
  attr1 attr2 attr3
id1   0   1   2
id2   3   4   5
 
# 宽表形式(dataframe)转变为堆叠形式（key-value）形式
# 数据库中常以该形式存储
df_key_value = df.stack().reset_index()
df_key_value.columns = [\'id\', \'attr\', \'value\']
print(df_key_value)
out:
id	attr	value
0	id1	attr1	0
1	id1	attr2	1
2	id1	attr3	2
3	id2	attr1	3
4	id2	attr2	4
5	id2	attr3	5
 
# 堆叠转换为宽表形式
 
# 用set_index创建层次化索引，在用unstack重塑
# unstack中作为旋转轴的变量（如attr），其值会作为列变量展开
df_key_value.set_index([\'id\',\'attr\']).unstack(\'attr\')
out:
value
attr	attr1	attr2	attr3
id			
id1	0	1	2
id2	3	4	5
 
# 多层索引转化为宽表
df_long = df_key_value.set_index([\'id\',\'attr\']).unstack(\'attr\')[\'value\'].reset_index()
df_long
out:
attr	id	attr1	attr2	attr3
0	id1	0	1	2
1	id2	3	4	5
 
# 堆叠转换为宽表的快捷键---pivot
df_key_value.pivot(\'id\',\'attr\',\'value\')
out:
attr	attr1	attr2	attr3
id			
id1	0	1	2
id2	3	4	5