python多进程map比apply快_python:map,apply,mapapply区别

apply:可作用与Series，也可以作用DataFrame

作用于Series

*data

CabinFareAge

0Acd7.829234.5

1Vafe7.000047.0

2Cfwrw9.687562.0

3Rfdf8.662527.0

4Mfdf12.287522.0

*data['cabin']=data['Cabin'].apply(lambda x:x[0])

*data['cabin']

0 A

1 V

2 C

3 R

4 M

作用与DataFrame

*data[['Fare','Age']].apply(lambda x:x.max()-x.min())

Fare 5.2875

Age 40.0000

map只可作用于Series

*data['cabin1'] = data['Cabin'].map(lambda x:x[0])

*data['cabin1']

0 A

1 V

2 C

3 R

4 M

*data[['Fare','Age']].map(lambda x:x.max()-x.min())

AttributeError: 'DataFrame' object has no attribute 'map'

#还可以用作替换

*x = pd.Series([0,1,2], index=['one', 'two', 'three'])

*y = pd.Series(['foo', 'bar', 'baz'], index=[0,1,2])

*x.map(y) #x的值与y的index 值相同，可将替换，不同，以NaN填充

one foo

two bar

three baz

dtype: object

#作用于合并

*x.map('you have {} pen'.format,na_action=None) #na_action 对NaN填充，'ignore'忽视NaN不填充

one you have 0 pen

two you have 1 pen

three you have 2 pen

applymap 部分行、列，对所有元素进行操作。

操作对象可以是DataFrame 或者 Series