assign

DataFrame.assign

_DataFrame.assign(kwargs)**
给DataFrame分配一个新列，返回一个新的_DataFrame对象，不仅包含新添加的列，还包含所有原始的列。重新分配的现有列将被覆盖。

举例：使用count列和price列的乘积作为total列

import pandas as pd
df = pd.DataFrame({'site':['google', 'baidu', 'wiki', 'pandas'],
                    'count':[18, 39, 22, 45],
                    'price': [1.0, 2.0, 3.0, 4.0]})
df = df.assign(total=df['count'] * df['price'])
------------------------------------
    site    count    price    total
0    google    18        1.0        18.0
1    baidu    39        2.0        78.0
2    wiki    22        3.0        66.0
3    pandas    45        4.0        180.0

举例：为原本的site列的所有元素重新赋值

import pandas as pd
df = pd.DataFrame({'site':['google', 'baidu', 'wiki', 'pandas'],
                    'count':[18, 39, 22, 45],
                    'price': [1.0, 2.0, 3.0, 4.0]})
df = df.assign(site=['google','google','google','google'])
-------------------------------
    site    count    price
0    google    18        1.0
1    google    39        2.0
2    google    22        3.0
3    google    45        4.0

举例：将Lambda表达式的结果作为total列添加

import pandas as pd
df = pd.DataFrame({'site':['google', 'baidu', 'wiki', 'pandas'],
                    'count':[18, 39, 22, 45],
                    'price': [1.0, 2.0, 3.0, 4.0]})
df = df.assign(total=lambda x: x['price'] / 2)
-------------------------------------
    site    count    price    total
0    google    18        1.0        0.5
1    baidu    39        2.0        1.0
2    wiki    22        3.0        1.5
3    pandas    45        4.0        2.0

举例：将Series对象作为新的列

import pandas as pd
df = pd.DataFrame({'site':['google', 'baidu', 'wiki', 'pandas']})
s1 = pd.Series([18, 39, 22, 45], name='count')
s2 = pd.Series([1, 2, 3, 4], name='price')
df.assign(count=s1, price=s2)
----------------------------------------------------------------
    site    count    price
0    google    18        1
1    baidu    39        2
2    wiki    22        3
3    pandas    45        4