检查Pandas DataFrame单元格是否包含某些字符串

Question

假设我有以下Pandas DataFrame:

         a        b             
0        NAN      BABA UN EQUITY
1        NAN      2018  
2        NAN      2017
3        NAN      2016
4        NAN      NAN
5        NAN      700 HK EQUITY
6        NAN      2018  
7        NAN      2017
8        NAN      2016
9        NAN      NAN 

我想检查列中的每个单元格b以查看它是否包含字符串EQUITY.如果是这样,我想替换列中的单元格a,下一行直到一行都是NAN前一个字符串,以获取编辑后的DataFrame,如下所示:

         a                 b             
0        NAN               BABA UN EQUITY
1        BABA UN EQUITY    2018  
2        BABA UN EQUITY    2017
3        BABA UN EQUITY    2016
4        NAN               NAN
5        NAN               700 HK EQUITY
6        700 HK EQUITY     2018  
7        700 HK EQUITY     2017
8        700 HK EQUITY     2016
9        NAN               NAN         

我的实际DataFrame比上面的大得多,但格式类似.我对Pandas很新,但我想我可以通过sheet.loc在循环中使用和替换单元格值来找出文本替换部分 .

但是,我无法弄清楚如何检查单元格是否包含EQUITY.这似乎str.contains是我应该使用的,但我不清楚如何做到这一点.

谢谢!

Answer 1

import numpy as np
import pandas as pd

df = pd.DataFrame({'a': ['NAN', 'NAN', 'NAN', 'NAN', 'NAN', 'NAN', 'NAN', 'NAN', 'NAN', 'NAN'],
 'b': ['BABA UN EQUITY', '2018', '2017', '2016', 'NAN', '700 HK EQUITY', '2018', '2017', '2016', 'NAN']})

# Make sure that all NaN values are `np.nan` not `'NAN'` (strings)
df = df.replace('NAN', np.nan)
mask = df['b'].str.contains(r'EQUITY', na=True)
df.loc[mask, 'a'] = df['b']
df['a'] = df['a'].ffill()
df.loc[mask, 'a'] = np.nan

产量

                a               b
0             NaN  BABA UN EQUITY
1  BABA UN EQUITY            2018
2  BABA UN EQUITY            2017
3  BABA UN EQUITY            2016
4             NaN             NaN
5             NaN   700 HK EQUITY
6   700 HK EQUITY            2018
7   700 HK EQUITY            2017
8   700 HK EQUITY            2016
9             NaN             NaN

上面稍微有点棘手的是如何mask定义.请注意,str.contains 返回的Series不仅包含True和False值,还包含NaN:

In [114]: df['b'].str.contains(r'EQUITY')
Out[114]: 
0     True
1    False
2    False
3    False
4      NaN
5     True
6    False
7    False
8    False
9      NaN
Name: b, dtype: object

str.contains(..., na=True)用于使NaNs被视为True:

In [116]: df['b'].str.contains(r'EQUITY', na=True)
Out[116]: 
0     True
1    False
2    False
3    False
4     True
5     True
6    False
7    False
8    False
9     True
Name: b, dtype: bool

一旦你有了mask这个想法很简单:将值复制b到a任何地方mask为True:

df.loc[mask, 'a'] = df['b']

向前填充NaN值a:

df['a'] = df['a'].ffill()

a用NaN 替换值,只要mask为True:

df.loc[mask, 'a'] = np.nan