Featran: A Scala feature transformation library for data science and machine learning

Feature engineering is the process of selecting and transforming properties of a data set to prepare for training a machine learning model, and is a vital component of successful ML systems. Often, this task involves writing cumbersome boilerplate and is usually coupled to a specific processing system. At Spotify, we built Featran to simplify this time-consuming task and support several processing frameworks under a uniform API, leveraging the powerful features and type-safety of Scala. This talk will begin with an overview of big data and ML at Spotify, and then we’ll dive into the design and implementation of Featran.

Session length

40 minutes

Language of the presentation

English

Target audience

Intermediate: Requires a basic knowledge of the area

Who is your session intended to

Big data and machine learning engineers, data scientists

Speaker

Andrew Martin (Data Infrastructure Engineer, Spotify)