I have two pandas columns that are supposed to represent the interactions of 2 types of chemicals. I want the 'user' type of ID to be in one column and the id2 column to only have the 'id/' ID types. This basically means just switch the first two columns for any rows that are not in this order. But the thing is, for each row where I switch the order, I also need to flip the True/False value for the third column and I have about 7 other columns that I want to stay the same. I have a small example below.

id1           id2            a_acting          othercol1   othercol2
user/123      id/308         True              ...       ...

id/328        user/219       True              ...       ...
 
user/321      id/328         False              ...       ...

user/328      id/321         False              ...       ...

desired

id1           id2           a_acting          othercol1   othercol2
user/123      id/308         True              ...       ...

user/219      id/328         False             ...       ...
 
user/321      id/328         False             ...       ...

user/328      id/321         False             ...       ...

CodePudding user response：

Use boolean indexing:

# identify incorrect rows
m = df['id1'].str.startswith('id')

# swap
df.loc[m, 'id1'], df.loc[m, 'id2'] = df.loc[m, 'id2'], df.loc[m, 'id1']

# flip boolean
df.loc[m, 'a_acting'] = ~df.loc[m, 'a_acting']