根据多行计算熊猫中的列

Question

I have a dataframe contains following columns:

1. Events
2. Transaction ID in which this event happened
3. The Previous transaction ID

I want to add a column to calculate if this event happened in the previous transaction or not

items = pd.DataFrame({'event':['A','B','B','A','C','C','C'],
                  'transaction_ID':[1,2,3,4,5,6,7],
                  'previous_trans':[2,3,5,7,4,1,6]})
items["Same_Event_in_prev_trans"]=0

The values of the "Same_Event_in_prev_trans" column should be 0 1 0 0 0 0 1

I am not sure how to do that without for loops.

Thanks.

Answer 1

You can use lambda to check for the event for previous trans.

items["Same_Event_in_prev_trans"]=(
    items.apply(lambda x: 1 if x.event==items.set_index('transaction_ID')
                                             .loc[x.previous_trans,'event'] 
                            else 0, axis=1)
    )



items
Out[239]: 
  event  previous_trans  transaction_ID  Same_Event_in_prev_trans
0     A               2               1                         0
1     B               3               2                         1
2     B               5               3                         0
3     A               7               4                         0
4     C               4               5                         0
5     C               1               6                         0
6     C               6               7                         1

Answer 2

Not entirely sure about the logic, but checking if each event's previous_trans is within the transaction_ID set seems to give the desired output:

items["Same_Event_in_prev_trans"] = (items.groupby('event', group_keys=False)
                                     .apply(lambda g: g.previous_trans.isin(g.transaction_ID))
                                     .astype(int))
​
items
#   event   previous_trans  transaction_ID  Same_Event_in_prev_trans
#0      A                2              1                       0
#1      B                3              2                       1
#2      B                5              3                       0
#3      A                7              4                       0
#4      C                4              5                       0
#5      C                1              6                       0
#6      C                6              7                       1

Answer 3

How about

>> items['prev_event'] = pd.merge(items, items[['event', 'transaction_ID']], 
>>                                left_on='previous_trans', 
>>                                right_on='transaction_ID')['event_y']
>> items['same_event'] = (items['event'] == items['prev_event']).astype(int)