Big Data Now: 2016 Edition

Chapter 5. Machine Learning: Models and Training

In this chapter, Mikio Braun looks at how data-driven recommendations are computed, how they are brought into production, and how they can add real business value. He goes on to explore broader questions such as what the interface between data science and engineering looks like. Michelle Casbon then discusses the technology stack used to perform natural language processing at startup Idibon, and some of the challenges they’ve tackled, such as combining Spark functionality with their unique NLP-specific code. Next, Ben Lorica offers techniques to address overfitting, hyperparameter tuning, and model interpretability. Finally, Marco Tulio Ribeiro, Sameer Singh, and Carlos Guestrin introduce local interpretable model-agnostic explanations (LIME), a technique to explain the predictions of any machine-learning classifier.

What Is Hardcore Data Science—in Practice?

By Mikio Braun

You can read this post on oreilly.com here.

During the past few years, data science has become widely accepted across a broad range of industries. Originally more of a research topic, data science has early roots in scientists’ efforts to understand human intelligence and to create artificial intelligence; it has since also proven that it can add real business value.

As an example, we can look at the company I work for—Zalando, one of Europe’s biggest fashion retailers—where data science is heavily used to provide data-driven recommendations, among other things. ...