Pandas dataframe cumprod by columns

Question

I have a daframe with yearly performance of three items. I want to add another column with the cumulated product by columns. The cumulated product of x is 0,42%. I tried with pandas cumprod either with axis=0 and 1 but it give me a error. There is a way to do this?

2016  2017  2018 
2,00% -3,00% 1,50%
4,00%  2,00% -1,00%
-5,00%  7,00% -2,50%

2016  2017  2018  TOTAL CUMULATED
2,00% -3,00% 1,50% 0,42%
4,00%  2,00% -1,00% 5,02%
-5,00%  7,00% -2,50% -0,89%

This is my code but it's not correct:

import pandas as pd
import numpy as np
data = [[0.02, -0.03, 1.50], [0.04, 0.02, -1.00], [-0.05, 0.07, -2.50]]
df = pd.DataFrame(data, columns=['2016', '2017', '2018'])
df['TOTAL CUMULATED'] = np.cumprod(1 + df) - 1

It give me:ValueError: Wrong number of items passed 3, placement implies 1

So, the code is not correct because it add a row at the end of the dataframe, instead i want to add another column with the cumprod of rows

Answer 1

Here it seems you want pandas.prod not np.cumprod . This should work:

df['TOTAL CUMULATED'] = (df+1).prod(axis=1) - 1

produces

      2016    2017    2018    TOTAL CUMULATED
--  ------  ------  ------  -----------------
 0    0.02   -0.03   0.015          0.004241
 1    0.04    0.02  -0.01           0.050192
 2   -0.05    0.07  -0.025         -0.0089125

Note I had to fix your df construction as the last column was off by x100