Welcome to Hands-On Machine Learning with R, your comprehensive guide to navigating the world of R Machine Learning. This book is designed to provide you with practical, hands-on modules covering a wide array of the most utilized machine learning techniques. Whether you’re interested in generalized low rank models, clustering algorithms, autoencoders, regularized models, random forests, gradient boosting machines, deep neural networks, or stacking/super learners, this book has you covered.
Inside, you’ll discover how to build and fine-tune these diverse models using R packages that are not only powerful but also known for their scalability. Our primary goal is to explain these techniques in a way that fosters a deep understanding of their strengths and weaknesses. We prioritize intuition development, keeping mathematical complexity to a minimum while providing ample resources for those who wish to delve deeper into the theoretical underpinnings.
Who Should Dive Into This Book?
This book is tailored to be a practical companion for anyone involved in the machine learning process. It serves as an excellent resource for learning about various approaches and gaining practical intuition about modern, powerful methods widely accepted in the machine learning community. If you’re already acquainted with analytic methodologies, this book will still prove invaluable as a reference for implementing these techniques using various R packages.
While the internet offers a plethora of videos, blog posts, and tutorials on machine learning, we noticed a gap in consistency, completeness, and an often biased approach towards specific packages. This realization sparked the creation of this book – to offer a balanced, comprehensive, and hands-on guide.
It’s important to note that this book is not an introductory text to R programming or general programming concepts. We assume you have a working knowledge of the R language, including function definition, object management, program flow control, and other fundamental tasks. If you’re new to R, we recommend starting with “R for Data Science” by Wickham and Grolemund 2016 to grasp the basics of data science with R, such as data importation, cleaning, transformation, visualization, and exploration. For those aiming to enhance their R programming skills, “Advanced R” by Wickham 2014 is an excellent resource.
Furthermore, this book is not designed for an in-depth theoretical exploration of machine learning algorithms. Several books already excel in this area, such as “Elements of Statistical Learning” (J. Friedman, Hastie, and Tibshirani 2001), “Computer Age Statistical Inference” (Efron and Hastie 2016), and “Deep Learning” (Goodfellow, Bengio, and Courville 2016)).
Instead, our focus is to empower R users to effectively utilize the machine learning toolkit within R. This involves leveraging various R packages like glmnet, h2o, ranger, xgboost, and lime to build models and extract meaningful insights from your data. We advocate for a hands-on learning approach, nurturing an intuitive understanding of machine learning through practical examples and just enough theory to solidify your knowledge. While reading this book without coding in R is possible, we strongly encourage you to actively engage with the provided code examples to maximize your learning experience.
Why R for Machine Learning?
Over the past two decades, R has become a leading tool for scientific computing and a consistent frontrunner in implementing statistical methodologies for data analysis. R’s prominence in data science is largely due to its extensive, vibrant, and ever-expanding ecosystem of third-party packages. Packages like tidyverse streamline common data analysis tasks, while h2o, ranger, xgboost, and others offer fast and scalable machine learning capabilities. For machine learning interpretability, packages such as iml, pdp, and vip are invaluable. Throughout this book, we will introduce you to many more tools that enhance your R machine learning workflow.
Conventions and Additional Resources
To help you navigate this book effectively, we use specific typographical conventions:
- strong italic: Denotes new terms.
- bold: Indicates package and file names.
inline code: Highlights functions or commands you can type directly.- Code chunks: Represent commands or text for user input.
<span>1</span> <span>+</span><span>2</span>
## [1] 3Look out for these visual cues within the text:
Signifies a tip or suggestion to enhance your understanding.
Highlights a general note or important information.
Indicates a warning or caution to be mindful of.
To further enrich your learning, we’ve included resources throughout the chapters that we’ve found incredibly useful for deeper exploration and practical application. Due to print limitations, the hard copy of this book provides a condensed version of the concepts and methods. However, extensive online supplementary material is available at https://koalaverse.github.io/homlr/. This online resource is continuously updated with extended chapter content (e.g., random forest package benchmarking) and new content (e.g., random hyperparameter search). You can also download the datasets used in the book, access teaching resources like slides and exercises, and much more.
Your Feedback is Welcome
We greatly value reader feedback. If you encounter any errors or bugs, please report them by posting an issue at https://github.com/koalaverse/homlr/issues.
Acknowledgements
We extend our heartfelt thanks to everyone who contributed feedback, typo corrections, and engaged in discussions during the book’s writing process. Our GitHub contributors include a long list of individuals who significantly improved this work. We also appreciate the invaluable input from colleagues and collaborators who enriched the machine learning content.
Software Environment
This book was developed using a specific software environment to ensure reproducibility. All code was executed on a 2017 MacBook Pro with detailed specifications listed in the original text, using the packages and R version detailed in the session information. This information is provided to give context to the computational environment in which the code examples were tested and run. You can refer to the original text for the complete list of packages and session details.
This book is your hands-on gateway to mastering R machine learning. We hope it empowers you to build robust models, gain insightful knowledge from your data, and confidently apply machine learning techniques in your projects.
