How to use FeatureTools to generate new features by crossing features in a table?

Question

Feature crossing is a very common technique to find the nonlinear relationships in a dataset. How to use FeatureTools to generate new features by crossing features in a table?

Answer 1

It is possible to cross every pair of numeric features automatically in Featuretools using the primitive Multiply. As a code example, suppose we have the fictional dataframe

       index  price  shares_bought       date
index                                        
1          1   1.00              3 2017-12-29
2          2   0.75              4 2017-12-30
3          3   0.60              5 2017-12-31
4          4   0.50             18 2018-01-01
5          5   1.00              1 2018-01-02

and we want to multiply price by shares_bought. We would run

es = ft.EntitySet('Transactions')
es.entity_from_dataframe(dataframe=df, entity_id='log', index='index', time_index='date')

from featuretools.primitives import Multiply

fm, features = ft.dfs(entityset=es,
                      target_entity='log',
                      trans_primitives=[Multiply])

to make the dataframe into an entityset, and then run DFS to apply the Multiply in all places possible. In this case, since there are only two numeric features, we'll get a feature matrix fm which looks like

       price  shares_bought  price * shares_bought
index                                             
1       1.00              3                    3.0
2       0.75              4                    3.0
3       0.60              5                    3.0
4       0.50             18                    9.0
5       1.00              1                    1.0

If we want to apply a primitive to a particular pair of features by hand, it is possible to do so using seed features. Our code would then be

n12_cross = Multiply(es['log']['price'], es['log']['shares_bought'])

fm, features = ft.dfs(entityset=es,
                      target_entity='log',
                      seed_features=[n12_cross])

to get the same feature matrix as above.

---

EDIT: To make the dataframe above, I used

import pandas as pd
import featuretools as ft
df = pd.DataFrame({'index': [1, 2, 3, 4, 5],
                   'shares_bought': [3, 4, 5, 18, 1],
                   'price': [1.00, 0.75, 0.60, 0.50, 1.00]})
df['date'] = pd.date_range('12/29/2017', periods=5, freq='D')