Level up your ML game with this tool!🚀

Data Science, AI and ML
1

Feature Engineering is a crucial step when it comes to building good Machine Learning models.

Feature Engineering is usually a manual process and might take some time to get insights required to make useful features.

And as most of the other processes in the ML pipeline, feature engineering is automated as well!

Enter FeatureTools.:zap:

Feature tools is an open source library meant specifically for automatic feature engineering.

The 3 main components of Feature Tools are:

  1. Entities
  2. Feature Primitives
  3. Deep Feature Synthesis (DFS)

Pretty confusing names. Let’s understand the gist of them.

:white_check_mark:An Entity is nothing but your DataFrame. A set of DataFrame becomes an Entity Set upon which Feature Tools will operate.

:white_check_mark:Feature Primitives are the methods applied to the entity relationships to build new features.

Methods such as aggregations like mean, median, sum, max. You can also define your custom primitives to build customised features.

:white_check_mark:Deep Feature Synthesis (DFS) is just a fancy name to the process by which Feature Tools actually does its business.

DFS takes in entities and feature primitives and generates a set of useful features.

What are your thoughts on FeatureTools?

#python #machinelearning #datascience