熊猫数据的使用功能转换

Question

我有一个日期，previous_day和价格的数据。

DATE       | prev_day |  price
01-01-2018 |    1     |  44
02-01-2018 |    1     |  45
02-01-2018 |    2     |  41
03-01-2018 |    1     |  54
03-01-2018 |    2     |  49
03-01-2018 |    3     |  46

我想使用下面的步骤创建第二天和前一天：

DATE       | prev_day | price | next_day | prior_day
01-01-2018 |    1     |  44   |          |    
02-01-2018 |    1     |  45   |          |    
02-01-2018 |    2     |  41   |          |    
03-01-2018 |    1     |  54   |          |    
03-01-2018 |    2     |  49   |          |
03-01-2018 |    3     |  46   |          |

其中由每个next_day组成的逻辑等于日期+1天和prev_day +1天的价格，prior_day等于日期-1天和prev_day -1天的价格。

Answer 1

你可以groupby "prev_day“和shift

df['prior_day'] = df.groupby('prev_day')['price'].shift()
df['next_day'] = df.groupby('prev_day')['price'].shift(-1)

产出：

         DATE  prev_day  price  prior_day  next_day
0  01-01-2018         1     44        NaN      45.0
1  02-01-2018         1     45       44.0      54.0
2  02-01-2018         2     41        NaN      49.0
3  03-01-2018         1     54       45.0       NaN
4  03-01-2018         2     49       41.0       NaN
5  03-01-2018         3     46        NaN       NaN