FeatureTools 介绍

Featuretools是一个执行自动特征工程的框架。它擅长于为深度学习把互相关联的数据集转换为特征矩阵。我们可以将特征构造的操作分为两类：「转换」和「聚合」。我们通过下面的例子来了解FeatureTools使用方法。
代码示例地址：
https://github.com/scottlinlin/auto_feature_demo.git

安装

pip install featuretools

快速入门

1、导入feauretool

import featuretools as ft

2、加载数据

#加载数据
clients = pd.read_csv('data/clients.csv', parse_dates = ['joined'])
loans = pd.read_csv('data/loans.csv', parse_dates = ['loan_start', 'loan_end'])
payments = pd.read_csv('data/payments.csv', parse_dates = ['payment_date'])

输出：

3、创建实体和实体集

#创建实体
es = ft.EntitySet(id = 'clients')

#添加clients实体
es = es.entity_from_dataframe(entity_id = 'clients', dataframe = clients, 
                              index = 'client_id', time_index = 'joined')

#添加loads实体
es = es.entity_from_dataframe(entity_id = 'loans', dataframe = loans, 
                              variable_types = {'repaid': ft.variable_types.Categorical},
                              index = 'loan_id', 
                              time_index = 'loan_start')


#添加pyments实体
es = es.entity_from_dataframe(entity_id = 'payments', 
                              dataframe = payments,
                              variable_types = {'missed': ft.variable_types.Categorical},
                              make_index = True,
                              index = 'payment_id',
                              time_index = 'payment_date')
#打印实体集
es

输出：

4、添加实体关系

# 通过client_id 关联clients和loans实体
r_client_previous = ft.Relationship(es['clients']['client_id'],
                                    es['loans']['client_id'])
es = es.add_relationship(r_client_previous)

# 通过loan_id 关联payments和loans实体
r_payments = ft.Relationship(es['loans']['loan_id'],
                             es['payments']['loan_id'])
es = es.add_relationship(r_payments)

#打印实体集
es

输出：

5、聚合特征,并生成新特征

#聚合特征,并生成新特征
features, feature_names = ft.dfs(entityset = es, target_entity = 'clients')
features.head()

输入：

6、聚合特征，通过指定聚合和转换函数生成新特征

#聚合特征，通过指定聚合agg_primitives和转换trans_primitives生成新特征
features, feature_names = ft.dfs(entityset = es, target_entity = 'clients', 
                                 agg_primitives = ['mean', 'max', 'percent_true', 'last'],
                                 trans_primitives = ['years', 'month', 'subtract', 'divide'])
features.head()

输出：