仅在某些行的列上应用pandas函数

Question

|   user  |       category    | val  | 
| ------  | ------------------| -----|
| user 1  | c1                |   3  |  
| user 1  | c2                |   4  |
| user 1  | c3                |   8  | 
| user 2  | c1                |   5  |
| user 2  | c2                |   9  | 
| user 2  | c3                |   10 |

比方说，我想将函数f应用于val列。返回数据帧的最佳方法是什么，例如，f仅应用于类别为c1的值？

|   user  |       category    | val     |
| ------  | ------------------| -----   | 
| user 1  | c1                |   f(3)  | 
| user 1  | c2                |   4     | 
| user 1  | c3                |   8     | 
| user 2  | c1                |  f(5)   | 
| user 2  | c2                |   9     |
| user 2  | c3                |   10    | 

Answer 1

使用numpy.where：

df['val'] = np.where(df['category'] == 'c1', df['val'].apply(f), df['val'])

或使用loc进行过滤：

mask = df['category'] == 'c1'
df.loc[mask, 'val'] = df.loc[mask, 'val'].apply(f)