book

Effective Data Science Infrastructure

Name: Effective Data Science Infrastructure
Author: Ville Tuulos
ISBN: 9781617299193

by Ville Tuulos

August 2022

Intermediate to advanced

352 pages

11h 36m

English

Manning Publications

Audiobook available

Read now

Unlock full access

inside front cover
Effective Data Science Infrastructure
Copyright
contents
front matter
forewordprefaceacknowledgmentsabout this bookWho should read this book?How this book is organized: A road mapAbout the codeliveBook discussion forumOther online resourcesabout the authorabout the cover illustration
1 Introducing data science infrastructure
1.1 Why data science infrastructure?1.1.1 The life cycle of a data science project1.2 What is data science infrastructure?1.2.1 The infrastructure stack for data science1.2.2 Supporting the full life cycle of a data science project1.2.3 One size doesn’t fit all1.3 Why good infrastructure matters1.3.1 Managing complexity1.3.2 Leveraging existing platforms1.4 Human-centric infrastructure1.4.1 Freedom and responsibility1.4.2 Data scientist autonomySummary
2 The toolchain of data science
2.1 Setting up a development environment2.1.1 Cloud account2.1.2 Data science workstation2.1.3 Notebooks2.1.4 Putting everything together2.2 Introducing workflows2.2.1 The basics of workflows2.2.2 Executing workflows2.2.3 The world of workflow frameworksSummary
3 Introducing Metaflow
3.1 The basics of Metaflow3.1.1 Installing Metaflow3.1.2 Writing a basic workflow3.1.3 Managing data flow in workflows3.1.4 Parameters3.2 Branching and merging3.2.1 Valid DAG structures3.2.2 Static branches3.2.3 Dynamic branches3.2.4 Controlling concurrency3.3 Metaflow in Action3.3.1 Starting a new project3.3.2 Accessing results with the Client API3.3.3 Debugging failures3.3.4 Finishing touchesSummary
4 Scaling with the compute layer
4.1 What is scalability?4.1.1 Scalability across the stack4.1.2 Culture of experimentation4.2 The compute layer4.2.1 Batch processing with containers4.2.2 Examples of compute layers4.3 The compute layer in Metaflow4.3.1 Configuring AWS Batch for Metaflow4.3.2 @batch and @resources decorators4.4 Handling failures4.4.1 Recovering from transient errors with @retry4.4.2 Killing zombies with @timeout4.4.3 The decorator of last resort: @catchSummary
5 Practicing scalability and performance
5.1 Starting simple: Vertical scalability5.1.1 Example: Clustering Yelp reviews5.1.2 Practicing vertical scalability5.1.3 Why vertical scalability?5.2 Practicing horizontal scalability5.2.1 Why horizontal scalability?5.2.2 Example: Hyperparameter search5.3 Practicing performance optimization5.3.1 Example: Computing a co-occurrence matrix5.3.2 Recipe for fast-enough workflowsSummary

6 Going to production
6.1 Stable workflow scheduling6.1.1 Centralized metadata6.1.2 Using AWS Step Functions with Metaflow6.1.3 Scheduling runs with @schedule6.2 Stable execution environments6.2.1 How Metaflow packages flows6.2.2 Why dependency managements matters6.2.3 Using the @conda decorator6.3 Stable operations6.3.1 Namespaces during prototyping6.3.2 Production namespaces6.3.3 Parallel deployments with @projectSummary
7 Processing data
7.1 Foundations of fast data7.1.1 Loading data from S37.1.2 Working with tabular data7.1.3 The in-memory data stack7.2 Interfacing with data infrastructure7.2.1 Modern data infrastructure7.2.2 Preparing datasets in SQL7.2.3 Distributed data processing7.3 From data to features7.3.1 Distinguishing facts and features7.3.2 Encoding featuresSummary
8 Using and operating models
8.1 Producing predictions8.1.1 Batch, streaming, and real-time predictions8.1.2 Example: Recommendation system8.1.3 Batch predictions8.1.4 Real-time predictionsSummary
9 Machine learning with the full stack
9.1 Pluggable feature encoders and models9.1.1 Developing a framework for pluggable components9.1.2 Executing feature encoders9.1.3 Benchmarking models9.2 Deep regression model9.2.1 Encoding input tensors9.2.2 Defining a deep regression model9.2.3 Training a deep regression model9.3 Summarizing lessons learnedSummary
Appendix A. Installing Conda
index
inside back cover

Content preview from Effective Data Science Infrastructure

8 Using and operating models

This chapter covers

Using machine-learning models to produce predictions that benefit real-world applications
Producing predictions as a batch workflow
Producing predictions as a real-time application

Why do businesses invest in data science applications? “To produce models” isn’t an adequate answer, because models are just bundles of data and code with no intrinsic value. To produce tangible value, applications must have a positive impact on the surrounding world. For instance, a recommendation model is useless in isolation, but when connected to a user interface, it can lower customer churn and increase long-term revenue. Or a model predicting credit risk becomes valuable when connected to a decision-support dashboard ...

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.

Read now

Unlock full access

More than 5,000 organizations count on O’Reilly

O’Reilly covers everything we've got, with content to help us build a world-class technology community, upgrade the capabilities and competencies of our teams, and improve overall team performance as well as their engagement.

Julian F.

Head of Cybersecurity

I wanted to learn C and C++, but it didn't click for me until I picked up an O'Reilly book. When I went on the O’Reilly platform, I was astonished to find all the books there, plus live events and sandboxes so you could play around with the technology.

Addison B.

Field Engineer

I’ve been on the O’Reilly platform for more than eight years. I use a couple of learning platforms, but I'm on O'Reilly more than anybody else. When you're there, you start learning. I'm never disappointed.

Amir M.

Data Platform Tech Lead

I'm always learning. So when I got on to O'Reilly, I was like a kid in a candy store. There are playlists. There are answers. There's on-demand training. It's worth its weight in gold, in terms of what it allows me to do.

Mark W.

Embedded Software Engineer

Architecting Data and Machine Learning Platforms

Publisher Resources

ISBN: 9781617299193Publisher Support Publisher Website Errata Page

Cloud Computing

Data Engineering

Data Science

AI & ML

Programming Languages

Software Architecture

IT/Ops

Security

Design

Business

Soft Skills

Effective Data Science Infrastructure

by Ville Tuulos

8 Using and operating models

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.