机器学习特征提取 | 自动特征工程featuretools

1、什么是Featuretools？

为了能使框架普适，就像pandas用于数据准备或scikit-learn用于机器学习。

链接：https://www.featuretools/

2、安装

通过源码安装，代码如下：

git clone https://github/featuretools/featuretools.git
cd featuretools
python setup.py install

通过pip安装，命令如下：

pip install featuretools

3、五分钟快速开始

1）首先导入相关包：

In [1]: import featuretools as ft

2）准备相关数据：

In [2]: data = ft.demo.load_mock_customer()
In [3]: customers_df = data["customers"]
In [5]: sessions_df = data["sessions"]
In [7]: transactions_df = data["transactions"]
In [10]: relationships = [("sessions", "session_id", "transactions", "session_id"),
   ....:                  ("customers", "customer_id", "sessions", "customer_id")]

3）特征综合：

In [11]: feature_matrix_customers, features_defs = ft.dfs(entities=entities,
   ....:                                                  relationships=relationships,
   ....:                                                  target_entity="customers")

In [12]: feature_matrix_customers