[Pandas] 数据迭代-CFANZ编程社区

1.迭代Series

Series本身是一个可迭代的对象，可直接对Series使用for语句来遍历它的值

import pandas as pd

df = pd.DataFrame([['liver','E',89,21,24,64],
                   ['Arry','C',36,37,37,57],
                   ['Ack','A',57,60,18,84],
                   ['Eorge','C',93,96,71,78],
                   ['Oah','D',65,49,61,86]
                  ], 
                   columns = ['name','team','Q1','Q2','Q3','Q4'])
# 迭代指定的列
for i in df.name:
    print(i)

# 效果和上面相同
# df.name.values返回array结构数据可用于迭代
for i in df.name.values:
    print(i)

迭代索引和指定的多列，使用python内置的zip函数将其打包为可迭代的zip对象

import pandas as pd

df = pd.DataFrame([['liver','E',89,21,24,64],
                   ['Arry','C',36,37,37,57],
                   ['Ack','A',57,60,18,84],
                   ['Eorge','C',93,96,71,78],
                   ['Oah','D',65,49,61,86]
                  ], 
                   columns = ['name','team','Q1','Q2','Q3','Q4'])

# 迭代索引和指定的两列
for i, n, q in zip(df.index, df.name, df.Q1):
    print(i, n, q)

2. df.iterrows()

df.iterrows()生成一个可迭代对象，将DataFrame行作为(索引，行数据)组成的Series数据对进行迭代。在for语句中需要两个变量来承接数据:一个为索引变量，即使索引在迭代中不会使用(这种情况可用useless作为变量名);另一个为数据变量，读取具体列时，可以使用字典的方法和对象属性的方法

df.iterrows()是最常用、最方便的按行迭代方法

import pandas as pd
df = pd.DataFrame([['liver','E',89,21,24,64],
                   ['Arry','C',36,37,37,57],
                   ['Ack','A',57,60,18,84],
                   ['Eorge','C',93,96,71,78],
                   ['Oah','D',65,49,61,86]
                  ], 
                   columns = ['name','team','Q1','Q2','Q3','Q4'])
# 迭代,使用name,Q1数据
for index, row in df.iterrows():
    print(index, row['name'], row.Q1)

3 df.itertuples()

df.itertuples()生成一个namedtuples类型数据，name默认名为Pandas，可以在参数中指定

与df.iterrows()相比,df.itertuples()运行速度会更快一些，推荐在数据量庞大的情况下优先使用

import pandas as pd
df = pd.DataFrame([['liver','E',89,21,24,64],
                   ['Arry','C',36,37,37,57],
                   ['Ack','A',57,60,18,84],
                   ['Eorge','C',93,96,71,78],
                   ['Oah','D',65,49,61,86]
                  ], 
                   columns = ['name','team','Q1','Q2','Q3','Q4'])
for row in df.itertuples():
    print(row)

以下是一些使用方法示例:

import pandas as pd

df = pd.DataFrame([['liver','E',89,21,24,64],
                   ['Arry','C',36,37,37,57],
                   ['Ack','A',57,60,18,84],
                   ['Eorge','C',93,96,71,78],
                   ['Oah','D',65,49,61,86]
                  ], 
                   columns = ['name','team','Q1','Q2','Q3','Q4'])

# 不包含索引数据
for row in df.itertuples(index=False):
    print(row)

# 自定义name
# namedtuples
for row in df.itertuples(index=False, name='Hudas'):
    print(row)

# 使用数据
for row in df.itertuples():
    print(row.Index, row.name)

4 df.items()

df.items()和df.iteritems()功能相同，它迭代时返回一个(列名，本列的Series结构数据),实现对列的迭代

如果需要对Series的数据再进行迭代，可嵌套for循环

import pandas as pd
df = pd.DataFrame([['liver','E',89,21,24,64],
                   ['Arry','C',36,37,37,57],
                   ['Ack','A',57,60,18,84],
                   ['Eorge','C',93,96,71,78],
                   ['Oah','D',65,49,61,86]
                  ], 
                   columns = ['name','team','Q1','Q2','Q3','Q4'])

# Series取前三个
for label, ser in df.items():
    print(label)
    print(ser[:3], end='\n\n')

5 按列迭代

除了df.items()，如需要迭代一个DataFrame的列，可以直接对DataFrame迭代，会循环得到列名

import pandas as pd

df = pd.DataFrame([['liver','E',89,21,24,64],
                   ['Arry','C',36,37,37,57],
                   ['Ack','A',57,60,18,84],
                   ['Eorge','C',93,96,71,78],
                   ['Oah','D',65,49,61,86]
                  ], 
                   columns = ['name','team','Q1','Q2','Q3','Q4'])

# 直接对DataFrame迭代
for column in df:
    print(column)

# 再利用df[列名]的方法迭代列
# 依次取出每个列
for column in df:
    print(df[column])

# 可对每个列的内容进行迭代:
for column in df:
    for i in df[column]:
        print(i)

# 可以迭代指定列
for i in df.name:
    print(i)

# 只迭代想要的列
l = ['name','Q1']
cols = df.columns.intersection(l)
for col in cols:
    print(col)