Transforming a list-like column

Rimodellare i dati con pandas

Maria Eugenia Inzaugarat

Data Scientist

List-like columns

DataFrame with a column containing lists

Transforming list-like columns

Arrow pointing from DataFrame to explode DataFrame

The .explode() method

Arrow pointing from DataFrame to explode DataFrame

The explode method

Exploding a column

cities

          city  country               zip_code
0  Los Angeles      USA  [90001, 90004, 90008]
1       Madrid    Spain  [28001, 28004, 28005]
2        Rabat  Morocco         [10010, 10170]

Exploding a column

cities_explode = cities['zip_code'].explode()
cities_explode

Exploding a column

cities[['city', 'country']]

Exploding a column

cities[['city', 'country']].merge(cities_explode,                                  )

Exploding a column

cities[['city', 'country']].merge(cities_explode, left_index=True, right_index=True)

          city  country zip_code
0  Los Angeles      USA    90001
0  Los Angeles      USA    90004
0  Los Angeles      USA    90008
1       Madrid    Spain    28001
1       Madrid    Spain    28004
1       Madrid    Spain    28005
2        Rabat  Morocco    10010
2        Rabat  Morocco    10170

Exploding a column in the DataFrame

cities_explode = cities.explode('zip_code')
cities_explode

          city  country zip_code
0  Los Angeles      USA    90001
0  Los Angeles      USA    90004
0  Los Angeles      USA    90008
1       Madrid    Spain    28001
1       Madrid    Spain    28004
1       Madrid    Spain    28005
2        Rabat  Morocco    10010
2        Rabat  Morocco    10170

Exploding a column in the DataFrame

cities_explode.reset_index(drop=True, inplace=True)

          city  country zip_code
0  Los Angeles      USA    90001
1  Los Angeles      USA    90004
2  Los Angeles      USA    90008
3       Madrid    Spain    28001
4       Madrid    Spain    28004
5       Madrid    Spain    28005
6        Rabat  Morocco    10010
7        Rabat  Morocco    10170

Empty lists

cities_new

          city  country               zip_code
0  Los Angeles      USA  [90001, 90004, 90008]
1       Madrid    Spain                     []
2        Rabat  Morocco         [10010, 10170]

cities_new.explode('zip_code')

          city  country zip_code
0  Los Angeles      USA    90001
0  Los Angeles      USA    90004
0  Los Angeles      USA    90008
1       Madrid    Spain      NaN
2        Rabat  Morocco    10010
2        Rabat  Morocco    10170

Chaining operations

cities

          city  country             zip_code
0  Los Angeles      USA  90001, 90004, 90008
1       Madrid    Spain  28001, 28004, 28005
2        Rabat  Morocco         10010, 10170

Chaining operations

cities['zip_code'].str.split(',', expand=True)

       0       1       2
0  90001   90004   90008
1  28001   28004   28005
2  10010   10170    None

Chaining operations

cites.assign(zip_code=                                )

Chaining operations

cites.assign(zip_code=cities['zip_code'].str.split(','))

Chaining operations

cites.assign(zip_code=cities['zip_code'].str.split(',')).explode('zip_code')

          city  country zip_code
0  Los Angeles      USA    90001
0  Los Angeles      USA    90004
0  Los Angeles      USA    90008
1       Madrid    Spain    28001
1       Madrid    Spain    28004
1       Madrid    Spain    28005
2        Rabat  Morocco    10010
2        Rabat  Morocco    10170

Let's practice!

Rimodellare i dati con pandas