如何根据不同的列值合并2行

Question

I'm new to Python. I'm using pandas and I have the below data with 3 fields Task , Status_From and Status_To as a dataframe.

If the Status_To of the first row is same as the Status_From of next row, then those 2 rows should be combined based on Task .

+------+-------------+-----------+
| Task | Status_From | Status_To |
+------+-------------+-----------+
| AAA  | 31-Aug-18   | 04-Sep-18 |
| BBB  | 21-Jun-18   | 21-Jun-18 |
| BBB  | 21-Jun-18   | 29-Jun-18 |
| BBB  | 29-Jun-18   | 29-Jun-18 |
| CCC  | 20-Aug-18   | 20-Aug-18 |
| CCC  | 24-Aug-18   | 24-Aug-18 |
| CCC  | 24-Aug-18   | 01-Sep-18 |
| DDD  | 06-Jul-18   | 06-Jul-18 |
| EEE  | 18-May-18   | 18-May-18 |
| FFF  | 01-Aug-18   | 01-Aug-18 |
| GGG  | 20-Apr-18   | 23-Apr-18 |
| GGG  | 23-Apr-18   | 23-Apr-18 |
| HHH  | 22-Jan-18   | 23-Jan-18 |
| HHH  | 23-Jan-18   | 23-Jan-18 |
| HHH  | 23-Jan-18   | 30-Jan-18 |
+------+-------------+-----------+

Output expected:

+------+-------------+-----------+
| Task | Status_From | Status_To |
+------+-------------+-----------+
| AAA  | 31-Aug-18   | 04-Sep-18 |
| BBB  | 21-Jun-18   | 29-Jun-18 |
| CCC  | 20-Aug-18   | 20-Aug-18 |
| CCC  | 24-Aug-18   | 01-Sep-18 |
| DDD  | 06-Jul-18   | 06-Jul-18 |
| EEE  | 18-May-18   | 18-May-18 |
| FFF  | 01-Aug-18   | 01-Aug-18 |
| GGG  | 20-Apr-18   | 23-Apr-18 |
| HHH  | 22-Jan-18   | 30-Jan-18 |
+------+-------------+-----------+

Tried with a 'for' loop and 'if' condition. But it didn't work. Is there a simple option to do this?

Answer 1

Assume your data already sorted, then you can use cumsum() to setup groups, find the last Status_To of each group and then drop_duplicates().

df1 = df.assign(
    g=df.groupby('Task').apply(lambda x: (x.Status_From != x.Status_To.shift()).cumsum()).reset_index(level=0, drop=True)
)

Output of df1 is:

#   Task Status_From  Status_To  g
#0   AAA   31-Aug-18  04-Sep-18  1
#1   BBB   21-Jun-18  21-Jun-18  1
#2   BBB   21-Jun-18  29-Jun-18  1
#3   BBB   29-Jun-18  29-Jun-18  1
#4   CCC   20-Aug-18  20-Aug-18  1
#5   CCC   24-Aug-18  24-Aug-18  2
#6   CCC   24-Aug-18  01-Sep-18  2
#7   DDD   06-Jul-18  06-Jul-18  1
#8   EEE   18-May-18  18-May-18  1
#9   FFF   01-Aug-18  01-Aug-18  1
#10  GGG   20-Apr-18  23-Apr-18  1
#11  GGG   23-Apr-18  23-Apr-18  1
#12  HHH   22-Jan-18  23-Jan-18  1
#13  HHH   23-Jan-18  23-Jan-18  1
#14  HHH   23-Jan-18  30-Jan-18  1

Then, use transform:

df1['Status_To'] = df1.groupby(['Task', 'g']).Status_To.transform('last')
df1 = df1.drop_duplicates(['Task','g']).drop('g', axis=1)

New output will be:

#   Task Status_From  Status_To
#0   AAA   31-Aug-18  04-Sep-18
#1   BBB   21-Jun-18  29-Jun-18
#4   CCC   20-Aug-18  20-Aug-18
#5   CCC   24-Aug-18  01-Sep-18
#7   DDD   06-Jul-18  06-Jul-18
#8   EEE   18-May-18  18-May-18
#9   FFF   01-Aug-18  01-Aug-18
#10  GGG   20-Apr-18  23-Apr-18
#12  HHH   22-Jan-18  30-Jan-18