How to Sum of previous three row values to Current Row in Python-CodePudding

How I can sum the previous three-row values and have that value in the current row of a new column in python.

You need to consider the IS_DEFAULT Column Sum.

DEFAULT_3_MONTHS is my new column name.

Input:

Input Image .

Expected output:

Expected Output

CodePudding user response：

You can use:

df['DEFAULT_3_MONTHS'] = (
    df.assign(IS_DEFAULT=df['IS_DEFAULT'].shift(fill_value=0))
      .groupby('ACCT_ID').rolling(3, min_periods=0)['IS_DEFAULT']
      .max().astype(int).droplevel(0)
)

Output:

>>> df
    ACCT_ID  IS_DEFAULT  DEFAULT_3_MONTHS
0   ABCV256           0                 0
1   ABCV256           0                 0
2   ABCV256           1                 0
3   ABCV256           0                 1
4   ABCV256           1                 1
5   ABCV256           1                 1
6   ABCV256           1                 1
7   ABCV256           0                 1
8   ABCV256           1                 1
9   ABCV256           1                 1
10  ABCV256           0                 1
11  ABCV256           1                 1
12  ABCV256           0                 1
13  ABCV256           1                 1
14  ABCV256           1                 1
15  ABCV256           1                 1
16  ABCV256           1                 1
17  ABCV256           1                 1
18  ABCV256           1                 1
19  ABCV256           1                 1
20  ABCV256           0                 1
21  ABCV256           0                 1
22  ABCV256           0                 1
23  ABCV256           1                 0
24  ABCV256           0                 1
25  ABCV256           1                 1
26  ABCV256           0                 1
27  ABCV256           0                 1
28  ABCV256           0                 1
29  ABCV256           0                 0
30  ABCV256           0                 0
31  ABCV256           0                 0
32  ABCV256           0                 0
33  ABCV256           0                 0
34  ABCV256           0                 0
35  ABCV256           0                 0

CodePudding user response：

If you are using pandas, I think this should work:

df["DEFAULT_3_MONTHS"] = df.IS_DEFAULT.shift().rolling(3, min_periods = 0).max()