数据清洗之高阶函数处理

青旅半醒 2023-05-22 06:59 121阅读 0赞

# 高阶函数处理 #

*  在dataframe中使用apply方法，调用自定义函数对数据进行处理
 *  函数apply，注意axis
 *  可以使用astype函数对数据进行转换
 *  可以使用map函数进行数据转换

import pandas as pd
    import numpy as np
    import os

os.getcwd()

'D:\\Jupyter\\notebook\\Python数据清洗实战\\数据清洗之数据转换'

os.chdir('D:\\Jupyter\\notebook\\Python数据清洗实战\\数据')

df = pd.read_csv('sam_tianchi_mum_baby.csv', dtype=str, encoding='utf-8')

df.head(5)

<table> 
 <thead> 
  <tr> 
   <th></th> 
   <th>user_id</th> 
   <th>birthday</th> 
   <th>gender</th> 
  </tr> 
 </thead> 
 <tbody> 
  <tr> 
   <th>0</th> 
   <td>2757</td> 
   <td>20130311</td> 
   <td>1</td> 
  </tr> 
  <tr> 
   <th>1</th> 
   <td>415971</td> 
   <td>20121111</td> 
   <td>0</td> 
  </tr> 
  <tr> 
   <th>2</th> 
   <td>1372572</td> 
   <td>20120130</td> 
   <td>1</td> 
  </tr> 
  <tr> 
   <th>3</th> 
   <td>10339332</td> 
   <td>20110910</td> 
   <td>0</td> 
  </tr> 
  <tr> 
   <th>4</th> 
   <td>10642245</td> 
   <td>20130213</td> 
   <td>0</td> 
  </tr> 
 </tbody> 
</table>

def f(x):
        if '0' in str(x):
            return '女'
        elif '1' in str(x):
            return '男'
        else:
            return '未知'

# apply函数可做很多其他处理
    df['性别'] = df['gender'].apply(f)

df.head(5)

<table> 
 <thead> 
  <tr> 
   <th></th> 
   <th>user_id</th> 
   <th>birthday</th> 
   <th>gender</th> 
   <th>性别</th> 
  </tr> 
 </thead> 
 <tbody> 
  <tr> 
   <th>0</th> 
   <td>2757</td> 
   <td>20130311</td> 
   <td>1</td> 
   <td>男</td> 
  </tr> 
  <tr> 
   <th>1</th> 
   <td>415971</td> 
   <td>20121111</td> 
   <td>0</td> 
   <td>女</td> 
  </tr> 
  <tr> 
   <th>2</th> 
   <td>1372572</td> 
   <td>20120130</td> 
   <td>1</td> 
   <td>男</td> 
  </tr> 
  <tr> 
   <th>3</th> 
   <td>10339332</td> 
   <td>20110910</td> 
   <td>0</td> 
   <td>女</td> 
  </tr> 
  <tr> 
   <th>4</th> 
   <td>10642245</td> 
   <td>20130213</td> 
   <td>0</td> 
   <td>女</td> 
  </tr> 
 </tbody> 
</table>

# 查看性别为未知数据
    df[df['gender'] == '2'].head(5)

<table> 
 <thead> 
  <tr> 
   <th></th> 
   <th>user_id</th> 
   <th>birthday</th> 
   <th>gender</th> 
   <th>性别</th> 
  </tr> 
 </thead> 
 <tbody> 
  <tr> 
   <th>46</th> 
   <td>49167150</td> 
   <td>20130818</td> 
   <td>2</td> 
   <td>未知</td> 
  </tr> 
  <tr> 
   <th>47</th> 
   <td>49983255</td> 
   <td>20140206</td> 
   <td>2</td> 
   <td>未知</td> 
  </tr> 
  <tr> 
   <th>51</th> 
   <td>52529655</td> 
   <td>20130611</td> 
   <td>2</td> 
   <td>未知</td> 
  </tr> 
  <tr> 
   <th>58</th> 
   <td>57711375</td> 
   <td>20130420</td> 
   <td>2</td> 
   <td>未知</td> 
  </tr> 
  <tr> 
   <th>106</th> 
   <td>99665637</td> 
   <td>20130926</td> 
   <td>2</td> 
   <td>未知</td> 
  </tr> 
 </tbody> 
</table>

del df['性别']

# map函数主要用于映射
    df['性别'] = df['gender'].map({ '0': '女性', '1':'男性', '2': '未知'})

df.head(5)

<table> 
 <thead> 
  <tr> 
   <th></th> 
   <th>user_id</th> 
   <th>birthday</th> 
   <th>gender</th> 
   <th>性别</th> 
  </tr> 
 </thead> 
 <tbody> 
  <tr> 
   <th>0</th> 
   <td>2757</td> 
   <td>20130311</td> 
   <td>1</td> 
   <td>男性</td> 
  </tr> 
  <tr> 
   <th>1</th> 
   <td>415971</td> 
   <td>20121111</td> 
   <td>0</td> 
   <td>女性</td> 
  </tr> 
  <tr> 
   <th>2</th> 
   <td>1372572</td> 
   <td>20120130</td> 
   <td>1</td> 
   <td>男性</td> 
  </tr> 
  <tr> 
   <th>3</th> 
   <td>10339332</td> 
   <td>20110910</td> 
   <td>0</td> 
   <td>女性</td> 
  </tr> 
  <tr> 
   <th>4</th> 
   <td>10642245</td> 
   <td>20130213</td> 
   <td>0</td> 
   <td>女性</td> 
  </tr> 
 </tbody> 
</table>

del df['性别']

# map函数也可传入自己定义的函数
    df['性别'] = df['gender'].map(f)

df.head(5)

# 脱敏处理
    # 可使用lambda函数
    df['user_id'].apply(lambda x: str(x).replace(x[1:3], '**')).head(5)

0        2**7
    1      4**971
    2     1**2572
    3    1**39332
    4    1**42245
    Name: user_id, dtype: object

df['birthday'].apply(lambda x: x[0:4]).head(5)

0    2013
    1    2012
    2    2012
    3    2011
    4    2013
    Name: birthday, dtype: object