Book Image

Test Driven Machine Learning

Book Image

Test Driven Machine Learning

Overview of this book

Related Content you might be interested in

Current Title:

Small book image
Test Driven Machine Learning
Nov, 2015 | 190 pages
No titles found
Table of Contents (16 chapters)
Test-Driven Machine Learning
Test-Driven Machine Learning
Credits
Credits
About the Author
About the Author
About the Reviewers
About the Reviewers
www.PacktPub.com
www.PacktPub.com
Preface
Preface
Free Chapter
1
Introducing Test-Driven Machine Learning
Introducing Test-Driven Machine Learning
Test-driven development
The TDD cycle
Behavior-driven development
Our first test
TDD applied to machine learning
Dealing with randomness
Different approaches to validating the improved models
Quantifying the classification models
Summary
2
Perceptively Testing a Perceptron
Perceptively Testing a Perceptron
Getting started
Summary
3
Exploring the Unknown with Multi-armed Bandits
Exploring the Unknown with Multi-armed Bandits
Understanding a bandit
Testing with simulation
Starting from scratch
Simulating real world situations
A randomized probability matching algorithm
A bootstrapping bandit
The problem with straight bootstrapping
Multi-armed armed bandit throw down
Summary
4
Predicting Values with Regression
Predicting Values with Regression
Refresher on advanced regression
Generating our own data
Building the foundations of our model
Cross-validating our model
Generating data
Summary
5
Making Decisions Black and White with Logistic Regression
Making Decisions Black and White with Logistic Regression
Generating logistic data
Measuring model accuracy
Generating a more complex example
Test driving our model
Summary
6
You're So Naïve, Bayes
You're So Naïve, Bayes
Gaussian classification by hand
Beginning the development
Summary
7
Optimizing by Choosing a New Algorithm
Optimizing by Choosing a New Algorithm
Upgrading the classifier
Applying our classifier
Upgrading to Random Forest
Summary
8
Exploring scikit-learn Test First
Exploring scikit-learn Test First
Test-driven design
Planning our journey
Getting choosey
Developing testable documentation
Summary
9
Bringing It All Together
Bringing It All Together
Starting at the highest level
The real world
What we've accomplished
Summary
Index
Index

About the Author

Justin Bozonier is a data scientist living in Chicago. He is currently a Senior Data Scientist at GrubHub. He has led the development of their custom analytics platform and also led the development of their first real time split test analysis platform which utilized Bayesian Statistics. In addition he has developed machine learning models for data mining as well as for prototyping product enhancements. Justin's software development expertise has earned him acknowledgements in the books Parallel Programming with Microsoft® .NET as well as Flow-Based Programming, Second Edition. He has also taught a workshop at PyData titled Simplified Statistics through Simulation.

His previous work experience includes being an Actuarial Systems Developer at Milliman, Inc., contracting as a Software Development Engineer II at Microsoft, and working as a Sr. Data Analyst and Lead Developer at Cheezburger Network amongst other experience.

Savannah Bozonier—the best partner I've ever had in life. Time and again she has made room in her life so I can push myself to do things that take an immense amount of time. Things like writing this book.

My friends and colleagues for their support and help which culminated in this book: Tom Hayden, Drew Fustin, and Andrew Slotnick.

My mentors across the years—Chad Boyer, Kelly Leahy, Robert Ream, James Thigpen, and Loren Bast.

My parents—I don't know what it's like to be told I can't do something. My life reflects that in every way.

Previous Section
Next Chapter