What’s with the “baby” in the name?
These lectures notes are largely based on a book called “Modern Dive,” hence the title “Baby Modern Dive.” I first stumbled upon the book when I was preparing for teaching this course in 2019. I personally benefited greatly from reading the book, and later teaching a course using its material. For many chapters, the lecture notes mirror the original book. Why creating this copy instead of just giving you the link to the original book? You ask.
Two reasons. First one is for the readers’ benefit. When I used Modern Dive for the first time while teaching this course in 2020 winter, the original book, although available online, was still in its draft mode. I was not completely satisfied with some aspects of it, so I made my own edits while developing the course material. Instead of dropping links everywhere in the lecture notes, I decided to integrate the original book (for the most part) with my edits (a small portion).
Second reason is more personal. Recreating the book, with some tinkering, has been a great learning experience for me. I won’t lie; it is a lot of fun.
What can you expect from this course
Experience the life cycle of a data analysis pipeline from raw data to meaninful narration, ready for readers of various background. This includes:
- explore, groom, and prepare data for further analysis, also known as data wrangling
- visualize data to inform model selection
- apply appropriate model to the data
- transform statistical output to prose, interpret results, make inferences
- develop a workflow using
Rthat ensures reproducible and shareable results
Woven throughout all of these action-packed “how” topics, are “why” topics. You will be exposed to statistical theories that could guide you in selecting models and making inferences. Spoiler alert! We will see formulas in this course. I do have some first-hand knowledge of the pain that the sight of formulars could inflict on some people, thanks to my degree in math cognition and years of teaching/coaching experience. If you are one of them, you are not alone. Fear not. This course is geared towards students who have very diverse backgrounds and therefore advanced math skill is not an requirement. Whenever a formula is introduced, I will unpack it so you can see how it works for yourself.
Why choosing R
In the pre-class survey, some of you mentioned that you had never heard of R. This section is included here for those of you who are new to R.
If you are expecting an objective recount of what R is and how wonderful it is, please refer to wikipedia and numerous books about R. What you will find here is my personal journey with R, an ulterly biased and opinionated version.
I heard about R for the first time at a Christmas party, perhaps 7 or 8 years ago. At the time, it was described as a geeky statistical software, powerful, but difficult to learn. My first instinct at hearing this was “I what to learn how to use R as well.” It was not until 2015, when I was starting to work on my PhD dissertation, that I actually decided to make a serious attempt. My motivation was very simple: I needed to run some analyses which I only knew then how to do in SAS. However, SAS being SAS, it did not support Mac. After wasting some time on finding a hacked solution to make SAS run on my Mac, I decided that it was time for me to adopt R. It would allow me to run the indended analysis properly. It is cross-platform. It is free. And what better opportunity to learn a new toy than a PhD dissertation?
Fast forward to 2017, I completed my dissertation, including all the analyses done with R. I would be lying if I said R was easy to learn. Did the effort pay off? That is a big fat yes.
Adopting R in my research opened up a new world for me. It taught me a new way of conducting research, one that is reproducible and shareable.
In the first ten years of my interaction with statistics, I was primarily using SPSS. And my routine can be summarized as:
import a dataset into SPSS, go through a series of clicking and dragging, copy and paste selected output into an editor, for example, MS word, fill in with texts and finish the report.
I had long felt uneasy about the workflow. However, it was not until I started transitioning to R that I could pinpoint the disadvantages in the previous process.
- The workflow is hardly reproducible because it involves a series of point-and-click, also known as GUI operations, or Graphical User Interface operations. It is already a struggle for the researcher themselves to remember the exact sequence of those clicking so as to generate the same output each time, let alone sharing such information with another person.
- Any change in the data source during analyses would require the authors to start from the top, to go through all the pointing and clicking again, sometimes worth hours of hardwork.
- It quickly become tedious, and more importantly, error-prone, when there are many figures and tables to copy and paste across documents.
This list could go on, but I think I have made my point clear. As you may have expected, once I started adopting R, I realized that I no longer had to deal with these problems. Let me be clear. It is not any intrisic feature of R, or its unique syntax that made R magical, or superior to SPSS and alike. The key lies in source code. When you work with R, you would spend most of your time writing and editing a source code, also known as a script, a file that is nothing more than a plain text, which contains detailed instructions about what operations to apply to the data, be it transposing, filtering, or aggregating. This source code can be saved on a computer and re-executed at any time, which would produce the same result as it did last time, provided that the source code has not been modified. It could also be shared with one another, producing the same results, transcending space and time change. Had to correct a typo in the data set? No problem. Go ahead and correct the typo, and re-run the source code and all your results will be updated accordingly.
Many advantageds of using R would apply to other script-based statistical programming languages. As we will find out later in this course, there are also some unique advantages of R, largely attributed to its vast user base, many of whom have been actively contributing to its improvement.
Who are the players?
By signing up for this course, you have agreed to be part of a one-semester-long game. Naturally, each of you is a player in this game. Unlike many other games, you are mostly competing with yourself in this one. And your score will reflect how much progress you make by the end of the semester compared to when you started.
Given that I made most of the rules for this game, I feel obliged to give you some background information about myself.
I wear many hats. I am an educator. I am a research psychologist. I am a human factors / user experience researcher. I am a statistics consultant. I worked with students advocating for their rights. I worked with seniors in the community. I am a geek.
My background is so mixed that I often have trouble telling people succinctly who I am. For a long time, I always felt a tingle of guilt for having so many identities, instead of a simple and unified identity. But now I think differently. What if I am all of them? Each of those identities will take the spotlight when the situation calls for it. In the context of this course, I am your instructor and cheerleader. I cannot force you to learn or to like this course. However, if you are willing to try, I will give you my full support.
Your teaching assistant
Unfortunately, we will not have a teaching assistant for this course this year, due to limited department resources. I will ask each of you please be patient if I am slightly behind the schedule.
Share a bit information about yourself with the rest of us.
I will continue to apply what worked last year, with some tweaking to adapt to the fully-online situation.
- Slack. For those of you who have not used slack before, it is an instant messaging tool, with some unique features tailored for workplace and classroom communications. Last semester, I created a slack workspace for the class, and it seemed to have worked. You will get a chance to familiarize yourself with the tool as part of your first assignment.
- cuLearn. I don’t expect we will use cuLearn for communication purposes, although it does have some neat functions such as anonymous message. If you have some constructive suggestion about the course but feel shy to talk to me, you can use this function. Without the help from a TA, I will only periodically monitor your posts on cuLearn.
- Email. I respect that some of you still prefer emails to other tools. Depending on the nature of your question, I may respond as soon as I see it, or wait a bit until I can process it in a quiet space so that I can respond in a thoughtful way. If you don’t hear from you in 48 hours, please send me a reminder. I would much appreciate it.
- One-on-one virtual meeting. Last year, this proved to be the most effective way of communication. I will provide more details on slack about how to book those meeetings with me.
Let’s face it.
Online class is not one of the most absorbing activities you have participated.
I took an online French class last fall and often caught myself multi-tasking during the class. I know the pain. The good news is that, in an online setting, your classmates cannot peek behind you and be distracted by the facebook page on your screen. However, I do encourage you to use some self-discipline and minimize distractions while you are attending the class. I understand that there will be times during the course when the content simply does not interest you, either because you already know the subject or my dull presentation. If you ever find yourself bored, please feel free to turn off the sound and/or video and spend you precious time on something more rewarding.
Structure of each lecture
Ideally, this course should be taught like workshops. Attendants learn by doing in each workshop, much like apprenticeship. Of course, you need to supplement with lectures on abstract theories from time to time, but the hands-on practice can be replaced with no amount of reading.
Hands-on workshps are best run in person, where the coach can pop over whenever a student encounters some difficuty. Although expected to be harder, I am still planning to run each lecture as workshops, with some anticipated overheads to overcome technical glitches.
Starting week 2, you will be expected to read the assigned material before coming to the lecture. The reading material will consist of notes on this website as well as occasional linked material. Based on the feedback from students last year, these lecture notes are fairly self-explanary. During the weekly lecture, we will spend time on addressing any questions you may have regarding the reading material, as well as on hands-on demonstration and practices.
To benefit from the lecture, it is essential that you finish the reading in advance. And after each lecture, it is recommended that you re-read the material to deepen your understanding.
Syllabus & assignments
Please see Appendix D for details.
Discuss in class
Getting help and learning more
This section is adapted from https://r4ds.had.co.nz/introduction.html#getting-help-and-learning-more
This book is not an island; there is no single resource that will allow you to master R. As you start to apply the techniques described in this book to your own data, you will soon find questions that we do not answer. This section describes a few tips on how to get help, and to help you keep learning.
If you get stuck, start with Google. Typically adding “R” to a query is enough to restrict it to relevant results: if the search isn’t useful, it often means that there aren’t any R-specific results available.
Google is particularly useful for error messages.
If you get an error message and you have no idea what it means,
try googling it!
Chances are that someone else has been confused by it in the past,
and there will be help somewhere on the web.
(If the error message isn’t in English, run
Sys.setenv(LANGUAGE = "en") and re-run the code;
you’re more likely to find help for English error messages.)
Among all the search results returned by Google, you will most likely encounter some from stackoverflow, a question and answer site for professional and enthusiast programmers who use programming languages such as Python and R. While most of the highly voted answers on stack overflow are trustworthy, do pay attention to when the answer was contributed. A solution that worked five years ago may no longer work for the same problem today, due to the fast evolving nature of R as a programming language. The bottom line: if you find multiple answers to the same question, prioritize the more recent ones.