使用Pandas GroupBy和value_counts找到最常见的值

Question

我正在处理一个表格中的两列。

+-------------+--------------------------------------------------------------+
|  Area Name  |                       Code Description                       |
+-------------+--------------------------------------------------------------+
| N Hollywood | VIOLATION OF RESTRAINING ORDER                               |
| N Hollywood | CRIMINAL THREATS - NO WEAPON DISPLAYED                       |
| N Hollywood | CRIMINAL THREATS - NO WEAPON DISPLAYED                       |
| N Hollywood | ASSAULT WITH DEADLY WEAPON, AGGRAVATED ASSAULT               |
| Southeast   | ASSAULT WITH DEADLY WEAPON, AGGRAVATED ASSAULT               |
| West Valley | CRIMINAL THREATS - NO WEAPON DISPLAYED                       |
| West Valley | CRIMINAL THREATS - NO WEAPON DISPLAYED                       |
| 77th Street | RAPE, FORCIBLE                                               |
| Foothill    | CRM AGNST CHLD (13 OR UNDER) (14-15 & SUSP 10 YRS OLDER)0060 |
| N Hollywood | VANDALISM - FELONY ($400 & OVER, ALL CHURCH VANDALISMS) 0114 |
+-------------+--------------------------------------------------------------+

我正在使用Groupby和value_counts来查找区域名称下的代码描述。

df.groupby(['Area Name'])['Code Description'].value_counts()

有没有办法只查看每个区域名称的前n个值？如果我在上面的代码中附加 .nlargest(3)，它只返回一个区域名称的结果。

+---------------------------------------------------------------------------------+
| Wilshire     SHOPLIFTING-GRAND THEFT ($950.01 & OVER)                         7 |
+---------------------------------------------------------------------------------+

- Patty Jula

---

我的问题更加清晰。 - Patty Jula

问题已重新开放，因为这里涉及到计算前N个值，而另一个问题是计算最常见的值，所以那里的答案不能用在这里。 - jezrael

Answer 1

使用value_counts的结果中的每个组都要使用head：

df.groupby('Area Name')['Code Description'].apply(lambda x: x.value_counts().head(3))

输出：

Area Name                                                                
77th Street  RAPE, FORCIBLE                                                  1
Foothill     CRM AGNST CHLD (13 OR UNDER) (14-15 & SUSP 10 YRS OLDER)0060    1
N Hollywood  CRIMINAL THREATS - NO WEAPON DISPLAYED                          2
             VIOLATION OF RESTRAINING ORDER                                  1
             ASSAULT WITH DEADLY WEAPON, AGGRAVATED ASSAULT                  1
Southeast    ASSAULT WITH DEADLY WEAPON, AGGRAVATED ASSAULT                  1
West Valley  CRIMINAL THREATS - NO WEAPON DISPLAYED                          2
Name: Code Description, dtype: int64

- Scott Boston

Answer 2

您可以执行双重分组：groupby。

s = df.groupby('Area Name')['Code Description'].value_counts()
res = s.groupby('Area Name').nlargest(3).reset_index(level=1, drop=True)

print(res)

Area Name    Code Description                                            
77th Street  RAPE, FORCIBLE                                                  1
Foothill     CRM AGNST CHLD (13 OR UNDER) (14-15 & SUSP 10 YRS OLDER)0060    1
N Hollywood  CRIMINAL THREATS - NO WEAPON DISPLAYED                          2
             ASSAULT WITH DEADLY WEAPON, AGGRAVATED ASSAULT                  1
             VANDALISM - FELONY ($400 & OVER, ALL CHURCH VANDALISMS) 0114    1
Southeast    ASSAULT WITH DEADLY WEAPON, AGGRAVATED ASSAULT                  1
West Valley  CRIMINAL THREATS - NO WEAPON DISPLAYED                          2
Name: Code Description, dtype: int64

- jpp