How do I get cumulative sum based on 'if' condition using a Boolean variable

Question

I have the following data frame:

vote_prprtn uniform_votesh  p_by_q    ln_p_by_q   p_ln_p_by_q   match
0.0116     0.009367        1.238434   0.213848   0.002481       False
0.0100     0.009367        1.067616   0.065428   0.000654       True
0.0065     0.009367        0.693950  -0.365355   -0.002375      True
0.0072     0.006000        1.200000   0.182321   0.001313       False
0.0048     0.006000        0.800000  -0.223143   -0.001071      True

I want to create another column which has the row value of p_ln_p_by_q when match = False and does a cumulative sum, row by row when match is True. This continues until the next False value. My final dataframe should look like this:

vote_prprtn uniform_votesh  p_by_q    ln_p_by_q   p_ln_p_by_q   match   final_val
0.0116     0.009367        1.238434   0.213848   0.002481       False    0.002481
0.0100     0.009367        1.067616   0.065428   0.000654       True     0.003135
0.0065     0.009367        0.693950  -0.365355   -0.002375      True     0.00076
0.0072     0.006000        1.200000   0.182321   0.001313       False    0.001313 
0.0048     0.006000        0.800000  -0.223143   -0.001071      True     0.000242

Any help on this would genuinely make me understand the parallels between excel and pandas.

Answer 1

groupby using a group defined by each time match==False then do the cumsum over each group.

df['final_val'] = df.groupby((~df.match).cumsum())['p_ln_p_by_q'].cumsum()

df['final_val']
# 0 0.002481
# 1 0.003135
# 2 0.000760
# 3 0.001313
# 4 0.000242

Answer 2

I don't know how your data frame is done, but you can try something in this style :

sum = 0
for row in frame.rows:
 if(row.match == False):
  sum = 0
 sum += row.p_ln_p_by_q
 row.final_val = sum

(Take care to adapt the way you read/write in your row, this is just a Python-based pseudo-code).