Chapter 1 Introduction

Preface

I developed this coursebook to help engineers begin to think about data. Most of my job at the university is to conduct research, yet most of the students who show up to my lab don’t know where to begin when presented with data. The irony is that, while engineering students are continuously drilled on how to solve problems, they are rarely taught how to seek them out. My undergraduate advisor, Dr. David Hemenway, once told me: “The data are always trying to tell you something.” This book is an introduction to data listening because listening comes before understanding.

In broad terms, scientific and engineering research is about discovery: finding out something new. In practice, however, research is really about failure. Over the short term, researchers fail in the act of discovery much more often than they succeed. Such failure is to be expected because cutting-edge research ain’t easy. Once you come to terms with accepting failure as a regular occurrence, you put yourself in a position to learn from it. To learn from failure requires that you observe and diagnose “what went wrong” and to do that, you need to listen to what the data are telling you. Let’s begin.

1.1 How to use this book

The coursebook is intended to be a sort of self-help guide for students who want to learn R programming and the art of engineering data science. The book is designed to get you started in the art, not master it. I’m not qualified to teach mastery in the art of R or engineering data science, so look elsewhere for that level of tutelage.

If you are new to these topics, you probably want to start at the beginning and proceed through each chapter sequentially. Some sections or material might seem boring or too easy. In that case, just skip to the end of the section and see if you can complete the exercises and answer the questions.

Nearly all of the graphics and data presented in this book were created or manipulated in R. In many places, however, I have hidden the code in order to streamline the message. If you ever wonder “How did he do that?” you can download any of the R Markdowns on the GitHub repository, where this coursebook is hosted.

1.2 Prerequisites

This course is intended for upper-level undergraduates who have completed MECH 231 (Experimental Design).

1.3 Grading

1.3.1 Grading for MECH 476

Course grades will be determined by the following three components:

Assessment component Percent of grade
Exams 60
Quizzes 15
Assignments 25

1.3.2 Homework submission

There will be a homework assignment per coursebook chapter. Each homework will be due at the start of the class period when lecture coverage of the next chapter commences. For example, Chapter 2 Homework will be due at the start of the first lecture regarding Chapter 3.

From RStudio, you will upload (“commit”/“push”) your homework to a private folder on our class GitHub Organization. Your commits will be time-stamped; any commit after the start of class on a given day will be considered late, and the assignment will be graded (or not graded) accordingly. Guidance on installing and connecting R/RStudio and Git/GitHub is provided in the next section.

1.4 Course set-up

This might be painful, but bear with me. There will be a lot of software development jargon (e.g., “commit”, “push”, “pull”), but the general idea is: We want to set up and learn how to use a collaborative tracking system. Git is a version control system (i.e., Word’s “track changes” feature on steroids). GitHub is the most common Git-based collaborative cloud (i.e., Dropbox on steroids). Everything in this class will be done within an RStudio Project on your local computer that is mirrored on a private GitHub repository. If you do not use GitHub (properly), you will receive an “incomplete” on the given assignment, so it is imperative that you take your time with these steps and read carefully. If you have any questions at each stage, ask!

The following guidance regarding R/RStudio and Git/GitHub draws heavily on Jenny Bryan’s book, Happy Git with R, and her related paper on version control. I sprinkled in some suggestions from others. If you would like to gain more background on basic Git and GitHub, take a look at these slides developed by Dr. Amelia McNamara.

1.4.1 Install R and RStudio

  1. Download and install the pre-compiled binary of the most recent version (4.4+) of R appropriate for your machine’s operating system
  2. Download and install the most recent, preview version of RStudio; then, navigate in RStudio to Tools > Global Options to NOT “Restore .RData into workspace at setup” and NEVER “save workspace to .RData on exit” (it is okay that you might not know what this means yet)
  3. If you’ve installed R and RStudio in the past, please download the latest versions and update the R packages with the following code:
# if you've previously installed R and RStudio, also update your R packages 
update.packages(ask = FALSE, checkBuilt = TRUE)
  1. If you want to generate PDF output from R Markdown documents, you will also need to install LaTex. I suggest taking the following approach in the RStudio Console, if you have never installed LaTex. More installation guidance can be found here.
# install R package
install.packages("tinytex")
# install LaTex "ingredients"
tinytex::install_tinytex()

1.4.2 Install Git and create a GitHub account

  1. Install Git. See here for OS-specific installation instructions. For Mac users, you need to install parts of Xcode for Mac OS and other things). For R 4.0+, this process is easier, as R now uses Apple Xcode 10.1 and GNU Fortran 8.2.

  2. Create a GitHub account. Pick a good user name!

  3. Introduce yourself to Git in RStudio with the following code in the Console. Provide your given name, not your user name, and the email address you used in creating your GitHub account. These commands return nothing, but you can check that it worked with git config --global --list in the shell.

# install `usethis` R package if needed (do this exactly once):
## install.packages("usethis")

library(usethis)
usethis::use_git_config(user.name = "John Doe", 
                        user.email = "john.doe@colostate.edu")

Note: Avoid committing credentials or other sensitive information to GitHub by “vaccinating” with usethis::git_vaccinate().

  1. Optional but recommended for new users: Consider downloading a GUI Git Client to make version control easier and to build intuition. GitHub Desktop is likely sufficient for this course, but the choice is yours.
  2. Optional: Sign up for free student perks via GitHub Education.

1.4.3 Connect Git, GitHub, and RStudio

  1. Read and follow the instructions in Chapters 9-13 exactly. I hope you won’t need to look at Chapter 14! Move slowly and carefully, and pay attention to the specific needs for your operating system.

1.4.4 Create new project, GitHub first

  1. Keep in mind: You should save the local R Project from this step in a top-level directory. We suggest creating an R folder at the top of your Documents folder (or OS-specific equivalent) to contain all of your R Projects. For the duration of this course, you should have a directory pathname like /user/Documents/R/[YourLastName]-MECH476. We will discuss directory structures and pathnames more in the next chapter.
  2. Read and follow these instructions exactly with the following additions:
  • Give your GitHub username to the Instructor, so you can be added to our GitHub Organization
  • Work with the Instructor to create a private repository within the organization, labeled [YourLastName]-MECH476
  • Confirm connection between your R Project and the GitHub repository, make subfolders (data, code, figs) within your [YourLastName]-MECH476 folder on your local drive, and push them up to GitHub; this supports a project-oriented workflow

1.4.5 Use RStudio/GitHub system for homework submission

For homework submission, you will download the R Markdown templates provided to you and save and edit them within your own R Project, which is connected a private repository on the class GitHub Organization. In order to track your changes and communicate with the Instructor, you will regularly commit changes to your R Project files with meaningful commit messages. We will practice commits, pushes, and pulls to the master branch (main copy of R Project) during class.

1.4.6 Confirm successful set-up

At this point, you should be able to commit, push to, and pull from the master branch of your private GitHub repository within the RStudio IDE. In later chapters, we will provide more information on these interfaces, and you will have plenty of opportunities to practice this workflow. For now, “minimally functional” is good enough!

Once you have successfully installed and connected R/RStudio and Git/GitHub, open an issue on YOUR private repository within the GitHub Organization. Mention/assign SmogDr to let me know everything is working properly, or to request more help.

Then, in the public repository for class-related questions and discussion, open another issue. You can ask a question, share any course-related concerns, or post a brief comment about what you hope to gain from this course. Remember to mention/assign SmogDr, so that I am alerted to your post.

1.4.7 Asking for help (properly)

All questions regarding technology and code should be directed to the Teaching Assistant via GitHub Issues on this repository. If the question requires you to include full code, please consider using the R package to generate reproducible examples: reprex. Please watch this tutorial on how to use reprex. Essentially, you are copy-pasting self-contained code in the GitHub Issue, so I can recreate your work and help you more effectively. You won’t do things perfectly as you start, but, hopefully by the end of the semester, you will have an efficient, reproducible workflow and effective solution-seeking toolbox.

1.5 Coursebook

This coursebook will serve as the only required textbook for this course. I regularly edit and add to this book, so content may change somewhat over the semester. We typically cover about a chapter of the book every 1-2 weeks of the course. You need to follow along and read this book thoroughly.

This coursebook includes:

  • Links to the slides presented in class for each topic
  • In-course exercises, typically including links to the data used in the exercise
  • Homework assignments
  • Appendix of reference distributions
  • A list of vocabulary and concepts that should be mastered for each quiz

If you find any typos or bugs, or if you have any suggestions for how the book can be improved, feel free to post it on the book’s GitHub Issues.

This book was developed using Yihui Xie’s bookdown framework. The book is built using code that combines R code, data, and text to create a book for which R code and examples can be re-executed every time the book is re-built, which helps identify bugs and broken code examples quickly. The online book is hosted using GitHub’s free GitHub Pages. All material for this book is available and can be explored at the book’s GitHub repository.

1.5.1 Other helpful books (not required)

The best book to supplement the coursebook and lectures for this course is R for Data Science by Garrett Grolemund and Hadley Wickham. The entire book is freely available online through the same format of the coursebook. You can also purchase a paper version of the book published by O’Reilly for around $40. This book is an excellent and up-to-date reference by some of the best R programmers in the world.

There are a number of other useful books available on general R programming, including:

The R programming language is used extensively within certain fields, including statistics and bioinformatics. If you are using R for a specific type of analysis, you will be able to find many books with advice on using R for both general and specific statistical analysis, including many available in print or online through the CSU library.

1.6 Acknowledgements

Most of the introductory material for this book was adapted from Dr. Brooke Anderson’s course on R Programming for Research , to whom I owe thanks not only for the materials but for the many helpful discussions. I would also like to acknowledge John Tukey, one of the pioneers of exploratory data analysis, and the creators of the NIST Engineering Statistics Handbook, from which I have drawn many techniques. Kathleen Wendt @Wendtke, the original TA for this course, also contributed a tremendous amount of effort to the original course setup and execution in 2020. Thank you!