python - 在循环python中重新定义变量

Question

使用此代码，

import numpy as np
import pandas as pd

df = pd.DataFrame ({'Date':['2000-01-01', '2000-02-01', '2000-03-01',
                            '2000-01-01', '2000-02-01', '2000-03-01','2000-04-01'
                           
                           ],
                  
                     'id':['1', '1', '1', '2', '2', '2','2'],
                     'bal_tot':[10, 20, 30, 40, 50, 60,70],
'bal_d1_pct': ['nan', 1, 2, 'nan', 3, 4, 5]
                                      
                    })

我需要为每个 id 创建一个新变量 'fore'。例如，

if Date = '2000-01-01' and id=1 then fore = 10
if Date = '2000-02-01' and id=1 then fore = 10 *(1+1/100) = 10.1
if Date = '2000-03-01' and id = 1 then fore = 10.1 * (1+2/100) = 10.302
etc.

最终数据应如下所示：

怎么做？

Answer 1

你可以做一个groupby，但你需要先让你bal_d1_pct成为数字：

df.bal_d1_pct = pd.to_numeric(df.bal_d1_pct, errors='coerce').fillna(0)/100 + 1

df['fore'] = (df.groupby('id')
   .apply(lambda x: x.bal_tot.iloc[0] * x.bal_d1_pct.cumprod())
   .reset_index('id',drop=True)
)

输出：

         Date id  bal_tot  bal_d1_pct     fore
0  2000-01-01  1       10        1.00  10.0000
1  2000-02-01  1       20        1.01  10.1000
2  2000-03-01  1       30        1.02  10.3020
3  2000-01-01  2       40        1.00  40.0000
4  2000-02-01  2       50        1.03  41.2000
5  2000-03-01  2       60        1.04  42.8480
6  2000-04-01  2       70        1.05  44.9904