create unique identifier in dataframe based on combination of columns, but only for duplicated rows

Question

A corollary of the question here: create unique identifier in dataframe based on combination of columns

In the foll. dataframe,

    id  Lat         Lon         Year    Area    State
50319   -36.0629    -62.3423    2019    90  Iowa
18873   -36.0629    -62.3423    2017    90  Iowa
18876   -36.0754    -62.327     2017    124 Illinois
18878   -36.0688    -62.3353    2017    138 Kansas

I want to create a new column which assigns a unique identifier based on whether the columns Lat, Lon and Area have the same values. Eg in this case rows 1 and 2 have the same values in those columns and will be given the same unique identifier 0_Iowa where Iowa comes from the State column. However, if there is no duplicate for a row, then I just want to use the state name. The end result should look like this:

id       Lat         Lon       Year    Area State   unique_id
50319   -36.0629    -62.3423    2019    90  Iowa    0_Iowa
18873   -36.0629    -62.3423    2017    90  Iowa    0_Iowa
18876   -36.0754    -62.327     2017    124 Illinois    Illinois
18878   -36.0688    -62.3353    2017    138 Kansas  Kansas

Answer 1

You can use an np.where :

df['unique_id'] = np.where(df.duplicated(['Lat','Lon'], keep=False), 
                            df.groupby(['Lat','Lon'], sort=False).ngroup().astype('str') + '_' + df['State'],
                           df['State'])

Or similar idea with pd.Series.where :

df['unique_id'] = (df.groupby(['Lat','Lon'], sort=False)
                     .ngroup().astype('str')
                     .add('_' + df['State'])
                     .where(df.duplicated(['Lat','Lon'], keep=False),
                            df['State']
                           )
                    )

Output:

      id      Lat      Lon  Year  Area     State unique_id
0  50319 -36.0629 -62.3423  2019    90      Iowa    0_Iowa
1  18873 -36.0629 -62.3423  2017    90      Iowa    0_Iowa
2  18876 -36.0754 -62.3270  2017   124  Illinois  Illinois
3  18878 -36.0688 -62.3353  2017   138    Kansas    Kansas