0

Here is a sample from my dataframe:

id      DPT_DATE  TRANCHE_NO  TRAIN_NO  J_X  RES_HOLD_IND
0     2017-04-01       330.0    1234.0 -1.0         100.0
1     2017-04-01       330.0    1234.0  0.0          80.0
2     2017-04-02       331.0    1235.0 -1.0          91.0
3     2017-04-02       331.0    1235.0  0.0          83.0
4     2017-04-03       332.0    1236.0 -1.0          92.0
5     2017-04-03       332.0    1236.0  0.0          81.0
6     2017-04-04       333.0    1237.0 -1.0          87.0
7     2017-04-04       333.0    1237.0  0.0          70.0
8     2017-04-05       334.0    1238.0 -1.0          93.0
9     2017-04-05       334.0    1238.0  0.0          90.0
10    2017-04-06       335.0    1239.0 -1.0          89.0
11    2017-04-06       335.0    1239.0  0.0          85.0
12    2017-04-07       336.0    1240.0 -1.0          82.0
13    2017-04-07       336.0    1240.0  0.0          76.0

This is a dataframe for Trains' reservation, DPT_DATE= date of departure TRAIN_NO= number of train J_X= days before departure (J_X=0.0 means the day of departure, J_X=-1 means day after departure) and RES_HOLD_IND is the reservation hold that day

I want to create a new column so for each DPT_DATE and TRAIN_NO gives me the RES_HOLD_IND for the day J_X=-1

Example (I want this):

id      DPT_DATE  TRANCHE_NO  TRAIN_NO  J_X  RES_HOLD_IND  RES_J-1
0     2017-04-01       330.0    1234.0 -1.0         100.0  100.0
1     2017-04-01       330.0    1234.0  0.0          80.0  100.0
2     2017-04-02       331.0    1235.0 -1.0          91.0  91.0
3     2017-04-02       331.0    1235.0  0.0          83.0  91.0
4     2017-04-03       332.0    1236.0 -1.0          92.0  92.0
5     2017-04-03       332.0    1236.0  0.0          81.0  92.0
6     2017-04-04       333.0    1237.0 -1.0          87.0  87.0
7     2017-04-04       333.0    1237.0  0.0          70.0  87.0

Thank you for your help!

phd
  • 82,685
  • 13
  • 120
  • 165
Oussama Jabri
  • 674
  • 1
  • 7
  • 18

1 Answers1

2

I think you need first filter by boolean indexing or query and then groupby with DataFrameGroupBy.ffill what works nice, if always -1 values are in first row per group:

df['RES_J-1'] = df.query('J_X == -1')['RES_HOLD_IND']
#alternative
#df['RES_J-1'] = df.loc[df['J_X'] == -1, 'RES_HOLD_IND']

df['RES_J-1'] = df.groupby(['DPT_DATE','TRAIN_NO'])['RES_J-1'].ffill()
print (df)
      DPT_DATE  TRANCHE_NO  TRAIN_NO  J_X  RES_HOLD_IND  RES_J-1
0   2017-04-01       330.0    1234.0 -1.0         100.0    100.0
1   2017-04-01       330.0    1234.0  0.0          80.0    100.0
2   2017-04-02       331.0    1235.0 -1.0          91.0     91.0
3   2017-04-02       331.0    1235.0  0.0          83.0     91.0
4   2017-04-03       332.0    1236.0 -1.0          92.0     92.0
5   2017-04-03       332.0    1236.0  0.0          81.0     92.0
6   2017-04-04       333.0    1237.0 -1.0          87.0     87.0
7   2017-04-04       333.0    1237.0  0.0          70.0     87.0
8   2017-04-05       334.0    1238.0 -1.0          93.0     93.0
9   2017-04-05       334.0    1238.0  0.0          90.0     93.0
10  2017-04-06       335.0    1239.0 -1.0          89.0     89.0
11  2017-04-06       335.0    1239.0  0.0          85.0     89.0
12  2017-04-07       336.0    1240.0 -1.0          82.0     82.0
13  2017-04-07       336.0    1240.0  0.0          76.0     82.0

If -1 is only one per group but not always first use:

df['RES_J-1'] = df.groupby(['DPT_DATE','TRAIN_NO'])['RES_J-1']
                  .apply(lambda x: x.ffill().bfill())
jezrael
  • 822,522
  • 95
  • 1,334
  • 1,252