Skip to content
Advertisement

How to fill a column with the sum of another column and the previous value of the same column?

I am building a financial model in Python. To do so, I need to calculate the “tax carry forward”. This position consists of the “EBT” plus the “tax carry forward” of the previous period. In the first period, the “tax carry forward” is equal to “EBT”.

I am currently trying to solve this with the df.shift() function:

df2["carry forward"] = df2["EBT"] + df2["carry forward"].shift(periods=1, fill_value=0)

This, however, doesn’t work properly. While I get correct results for the first two iterations (2021, 2022), it doesn’t work anymore from the third iteration onwards.

       EBT           carry forward  
year                                                                            
2021 -377893.353711 -377893.353711  
2022 -282754.978037 -660648.331748 
2023 -224512.990469 -507267.968506  
2024 -167696.637680 -392209.628149

The carry forward, as shown in the table above, for the year 2023 is the sum of EBT 2023 and 2022, which is incorrect. I can’t quite figure out my mistake, because I am not sure how exactly Python is populating columns in a dataframe. To me, it looks like Python isn’t populating the dataframe row by row, but rather simultaneously. Is this the problem? If so, how do I work around it? Is there a better way to do the task than the df.shift() function?

Advertisement

Answer

I assume that you are actually looking for a cumulative sum of the column EBT for the column of carry_forward?

For the input of:

                EBT  carry_forward
year                              
2021 -377893.353711              0
2022 -282754.978037              0
2023 -224512.990469              0
2024 -167696.637680              0

with:

df["carry_forward"] = df["EBT"].cumsum()
df

You will get:

                EBT  carry_forward
year                              
2021 -377893.353711  -3.778934e+05
2022 -282754.978037  -6.606483e+05
2023 -224512.990469  -8.851613e+05
2024 -167696.637680  -1.052858e+06
User contributions licensed under: CC BY-SA
8 People found this is helpful
Advertisement