0

Let's say I've the below data frame:

df = pd.DataFrame(
    np.random.randn(10, 3),
    index=pd.date_range("1/1/2022", freq="D", periods=10),
    columns=["A", "B", "C"],
    )

df

enter image description here

I want to add column D, so that we check the value of B at each date, if it is positive then D equals B of the previous day, if it is negative, then it is max of A and C of the previous day.

For example:

  1. At 2022-01-05, B is positive, then D should be equal to the value of B at day 2022-01-04 that is -1.329675
  2. At 2022-01-03, B is negative, then D should be equal to the max of A and C at day 2022-01-02 that is 0.562377
Hasan A Yousef
  • 22,789
  • 24
  • 132
  • 203
  • Perhaps this is helpful: https://stackoverflow.com/questions/19913659/pandas-conditional-creation-of-a-series-dataframe-column – le_camerone Aug 08 '22 at 05:42

1 Answers1

2

You can try shift with np.where

df['D'] = np.where(df['B'] > 0,
                   df['B'].shift(),
                   df[['A', 'C']].shift().max(axis=1))

print(df)

                   A         B         C         D
2022-01-01  1.495712  0.268461 -0.954359       NaN
2022-01-02 -0.585978 -0.214993  1.110485  1.495712
2022-01-03 -0.038061 -0.779959 -0.445906  1.110485
2022-01-04 -1.145300  0.436302 -0.480348 -0.779959
2022-01-05 -0.297486 -0.013070 -0.416821 -0.480348
2022-01-06  0.244772  0.288131  0.439024 -0.013070
2022-01-07  2.887469  0.608421 -0.432701  0.288131
2022-01-08  0.414783 -0.031525 -2.629204  2.887469
2022-01-09  0.786881  0.505138 -0.817680 -0.031525
2022-01-10 -1.097626  0.750188  1.216392  0.505138
Ynjxsjmh
  • 28,441
  • 6
  • 34
  • 52
  • How does this work in terms of the second condition `df['B'] < 0`? Is it like a if and elseif situation? – le_camerone Aug 08 '22 at 05:50
  • 1
    @le_camerone It's like a if and elseif statement, but since OP doesn't include the condition zero, I classify it to the situation `df['B'] < 0`. – Ynjxsjmh Aug 08 '22 at 05:54