How can I get back real value from cumulative sum in pandas

Question

Suppose, I've a DataFrame df.

>>> DATA = {'id':[1,2,3,4,5], 'salary':[1200,2300,2400,3620,2100] } 
>>> df = DataFrame(DATA)
>>> df
   id  salary
0   1    1200
1   2    2300
2   3    2400
3   4    3620
4   5    2100

From this DataFrame df, I can get a new DataFrame df1 with cumulative sum of salary

>>> df['salary'] = df['salary'].cumsum() 
>>> df
   id  salary
0   1    1200
1   2    3500
2   3    5900
3   4    9520
4   5   11620

This is very common scenario.

Now, What about if I am given df1 and I have to find df.

   id  salary            id  salary
0   1    1200             0   1    1200
1   2    3500             1   2    2300
2   3    5900     ==>     2   3    2400
3   4    9520             3   4    3620
4   5   11620             4   5    2100

All I have to find actual salary for all id from its cumulative sum.

Answer 1

>>> df
   id  salary
0   1    1200
1   2    3500
2   3    5900
3   4    9520
4   5   11620
>>> df['salary'] = df['salary'].diff().fillna(df['salary'].iloc[0])
>>> df
   id  salary
0   1    1200
1   2    2300
2   3    2400
3   4    3620
4   5    2100

although .fillna is not efficient in here, because it is only the first value which comes out null. so, you just need to replace the first value with cumulative value at .iloc[0].