基于DataFram中的另一列填充缺少的值

Question

我正在使用Pandas和numpy，对于下面的数据框，让我们称之为'data'，对于data['Borough']=='Unspecified'的Borough值，我需要使用它左侧的Incident zip字段中的邮政编码来查找匹配的邮政编码和Borough的Incident zip列。一旦找到了，就应该用自治区的名称替换“Unspecified”。下面是测试链接：https://colab.research.google.com/drive/1PgPbS7KxOrNfok3jtKoC55vXAXzK2E_N#scrollTo=poYboz-jhRCN 单击Runtime->；Runall

Created Date               Complaint Type   Incident Zip    Borough
0   09/14/2017 04:40:33 PM  New Tree Request    11374       QUEENS
1   03/18/2017 10:09:57 AM  General Construc    11420       QUEENS
2   03/29/2017 12:38:28 PM  General Construc    11230       Unspecified
3   06/05/2017 12:47:55 PM  New Tree Request    10028       Unspecified
4   06/15/2017 11:56:36 AM  Dead/Dying Tree     10467       BRONX
5   06/19/2017 10:30:46 AM  New Tree Request    11230       MANHATTAN
6   06/29/2017 09:25:59 AM  New Tree Request    10028       MANHATTAN
7   07/01/2017 12:23:20 PM  Damaged Tree        10467       BRONX
8   07/01/2017 11:47:03 AM  Damaged Tree        10467       BRONX
9   07/10/2017 10:27:37 AM  General Construc    11385       QUEENS
10  07/13/2017 09:20:53 PM  General Construc    11385       QUEENS

Answer 1

或者：

df.Borough.replace('Unspecified',np.nan,inplace=True)
df.Borough = df.sort_values(by='Incident Zip').groupby('Incident Zip')['Borough'].apply(lambda x : x.ffill().bfill())
>>df



             Created Date   Complaint Type   Incident Zip    Borough
0   09/14/2017 04:40:33 PM  New Tree Request         11374     QUEENS
1   03/18/2017 10:09:57 AM  General Construc         11420     QUEENS
2   03/29/2017 12:38:28 PM  General Construc         11230  MANHATTAN
3      2017-05-06 12:47:55  New Tree Request         10028  MANHATTAN
4   06/15/2017 11:56:36 AM   Dead/Dying Tree         10467      BRONX
5   06/19/2017 10:30:46 AM  New Tree Request         11230  MANHATTAN
6   06/29/2017 09:25:59 AM  New Tree Request         10028  MANHATTAN
7      2017-01-07 12:23:20      Damaged Tree         10467      BRONX
8      2017-01-07 11:47:03      Damaged Tree         10467      BRONX
9      2017-10-07 10:27:37  General Construc         11385     QUEENS
10  07/13/2017 09:20:53 PM  General Construc         11385     QUEENS

Answer 2

IIUC，您希望使用数据帧中的其他值来填充缺少的值。你可以用map来做这件事。你知道吗

首先，生成一系列映射到自治区的邮政编码。你知道吗

mapping = (df.query('Borough != "Unspecified"')
             .drop_duplicates('Incident Zip')
             .set_index('Incident Zip')
             .Borough)
mapping

Incident Zip
11374       QUEENS
11420       QUEENS
10467        BRONX
11230    MANHATTAN
10028    MANHATTAN
11385       QUEENS
Name: Borough, dtype: object

现在，将其传递给map，并使用fillna将未映射的值填充为“Unspecified”。你知道吗

df['Borough'] = df['Incident Zip'].map(mapping).fillna('Unspecified')

df
             Created Date    Complaint Type  Incident Zip    Borough
0  09/14/2017 04:40:33 PM  New Tree Request         11374     QUEENS
1  03/18/2017 10:09:57 AM  General Construc         11420     QUEENS
2  03/29/2017 12:38:28 PM  General Construc         11230  MANHATTAN
3  06/05/2017 12:47:55 PM  New Tree Request         10028  MANHATTAN
4  06/15/2017 11:56:36 AM   Dead/Dying Tree         10467      BRONX
5  06/19/2017 10:30:46 AM  New Tree Request         11230  MANHATTAN
6  06/29/2017 09:25:59 AM  New Tree Request         10028  MANHATTAN
7  07/01/2017 12:23:20 PM      Damaged Tree         10467      BRONX
8  07/01/2017 11:47:03 AM      Damaged Tree         10467      BRONX
9  07/10/2017 10:27:37 AM  General Construc         11385     QUEENS