Education on R Views

Learn and Teach R

Wed, 02 Dec 2020 00:00:00 +0000

If you haven’t explored the RStudio website in a while, your next visit may include a pleasant surprise. I recently went to the Tidymodels page, just to see what was new and was immediately drawn into the new landscape imagined by the RStudio education team. Clicking on Get Started I came to a fork and a choice of going farther with Tidymodels or backing up a bit. I went down the beginners path Finding Your Way to R and found myself in a well-lit wood, and I was not lost.

Like a park with well-marked trails, the Learning R section of the RStudio Education site branches off to R excursions graded to match the “hikers” experience. The Beginners trail starts in a safe place that should make even the absolutely terrified feel comfortable. It offers different modes of learning including videos, tutorials, books and even excursions to trusted third party sites that have their own feel.

The Intermediates section emphasizes learning how to get help, suggests some basic tools that should be useful no matter what path you select, and then points to places where you may already know you want to go: bioconductor, financial models or Spark clusters for example. The basic idea is that at this stage you know enough R to get some real work done, and a good path to making further progress might be to follow what interests you.

The Experts trail is for the intrepid who are ready to venture past terra firma and take a deep dive

into the foundations of R, or package building, or using Python, or exploring deep learning with Tensorflow. As with the others, this trail is well marked and offers tools and suggestions for making progress.

Perhaps the most pleasant surprise for an educator coming to the RStudio Education site is to see that the trails don’t stop with Expert. There is an entire section of the “park” marked off for teaching with trails that offer essential material for supporting both educators and online education. Learn to teach offers advice on how to develop as a teacher based on practical experience with the Carpentries. The section on Materials for teaching offers a panpoply of courses and workshops developed at RStudio that teachers can freely adapt to their needs. There is material here relevant to data wrangling, data science, R, tidyverse, shiny and more. The third section, Tools for teaching describes RStudio Cloud and other RStudio tools for creating a modern, interactive teaching infrastructure.

Whether you are just thinking about learning R or about to teach an advanced course, I think you might enjoy walking around the RStudio Education website. Maybe, I’ll get to Tidymodels next time.

Open-Source Authorship of Data Science in Education Using R

Wed, 01 Jul 2020 00:00:00 +0000

Joshua M. Rosenberg, Ph.D., is Assistant Professor of STEM Education at the University of Tennessee, Knoxville.

In earlier posts, we shared how we wrote Data Science in Education Using R as an open book (Post 1, Post 2). In this post, we describe what we consider to be the open-source authorship process we took to write the book.

We think of open-source authorship as a broader—and perhaps better—term for describing what authors of some open books undertake. In our characterization, open-source authorship draws upon:

parts of open-source software (OSS) values and tools
parts of open science that establish the importance of scholarly work beyond original, discovery research
the values surrounding the creation of open educational resources (OER)

We believe that combining elements from OSS, open science, and OER is notable because while OSS and open science emphasize the sharing of technical work (including technology and code) and OER emphasizes the sharing of resources, technical books have not been as much a focus of the conversation. Moreover, the way in which the conversation about open books has taken place in different communities and contexts means that some books that are open do not fully receive the attention (for being openly available) that they merit from those interested in OER. This also might mean that those involved with OSS development and open science may fail to recognize the creation of a book as a substantial contribution.

In this way, we argue for open-source authorship as an important, new type of work, one that we increasingly see by the authors of other books, especially in the R community¹ ² ³ ⁴.

After describing how we wrote our book in an open way, we elaborate on these ideas and draw connections to the process we undertook.

How We Wrote Data Science in Education Using R as an Open Book

Early in our process, we determined that we wanted to share the book in an open way. Since we were using GitHub as a repository for the book, it was easy for the contents of the book to be available for anyone to view–even before and as the book was being written. Despite the benefits of using GitHub, GitHub can be difficult to navigate for those who are unfamiliar with it, and so sharing the book in a more widely-accessible way was also important. To do this, we used {bookdown} and Netlify to share the book as a website. Additionally, we chose an easy-to-remember URL (http://datascienceineducation.com/) to help others (and us!) to be able to access it easily.

Being available for others to contribute was important. Because we used GitHub, we were able to receive feedback at a very early-stage on issues such as how we referred to data (as data or datum). Other issues (by non-authors) raised questions about whether certain content was in scope—such as content on gradebooks, which we included a chapter on. We found that apart from the five of us as authors, fifteen individuals made contributions, and another one hundred forty-four individuals starred the repository⁵. Moreover, we received feedback through Twitter and an email account we created for the book for those unfamiliar with GitHub (or Twitter) to be able to provide feedback directly to us. In this way, making the book available to others to contribute made the book better, and points to the importance of sharing work at only one stage of the writing process.

Lastly, we shared products that could be seen as tangential to the book, but which were important given its focus on data science and R. Namely, we created an R package, {dataedu}, to accompany the book. This package includes code to install the packages necessary to reproduce the book as well as all of the data sets used in it. By doing so, we invited others to contribute to the book in ways not related to its prose. This also led to (pleasantly) surprising contributions, including the creation of an iPython Notebook with python code to carry out comparable steps as those carried in a walkthrough chapter of our book.

Collectively, these practices—involving not only making the book open, but also planning for others to contribute and creating other, shared (open) products— comprise what we think of as the results of open-source authorship.

Originally a niche effort, open-source software (OSS) and OSS development are (likely not to the surprise of R users!) now widespread ⁶. There are some insights that can be gained from efforts to understand how OSS development proceeds. For example, in foundational, work Mockus et al. found that OSS is often characterized by a core group of 10-15 individuals contributing around 80% of the code, but that a group around one order of magnitude larger than that core will repair specific problems, and a group another order of magnitude larger will report issues⁷; proportions (generally) similar to those we found for those who contributed to our book.

Second, open science is both a perspective about how science should operate and a set of practices that reflect a perspective about how science should proceed ⁸ ⁹. Related to open science are open scholarly practices. Others trace the origin of the idea of open scholarly practices to a book by Boyer, who shared a broad description of intellectual (especially academic) work. This suggests that scholarly work is not only original, discovery research; it also includes the applications of advances in one’s own discipline (or “translational research”) and sharing the results of research with multiple stakeholders. Open science and open scholarly practices point to the scientific or scholarly contributions of open books; while different from original, scientific research, books such as our own—which focused on providing a language for data science in education—may serve as helpful examples (of open science) or forms of a broader view of scholarship.

Last, OER are “teaching, learning, and research resources that reside in the public domain or have been released under an intellectual property license that permits their free use and re-purposing by others” ¹⁰. These resources range from courses and books to tests and technologies. By being open, they are not only available to others to use, but also to reuse, redistribute (or share), revise (adapt or change the work), and remix (combining existing resources to create a new one) ¹¹. OER can serve as an inspiration for authors of open books, especially those who see their books as being used to teach and learn from. At the moment, OER and traditional publishing modes are largely separate: For most books that are published, the publisher retains the copyright, and authors are typically not allowed to share their book in the open, though this may be changing. Many authors of books about R have negotiated with their publisher to share their books in the open (often only as a website, as we have) in addition to sharing them through print and e-book formats. In addition, a number of platforms for creating books that are OER are emerging; one example is EdTech Books. There are increasing conversations related to making materials, resources, and even education as an enterprise more open; OER may be an area in which authors of books about R and other technical books can both learn from the work of authors as well as advance the conversation.

fin

This post was an effort to step back from what we did to write our book to reflect on what we meant by open-source authorship and to attempt to situate what we did (and what others have done) in broader conversations about OSS, open science, and OER. In this open mode, we invite others to revise or remix these ideas to advance other, new forms of authorship of books.

You can reach us on Twitter: Emily @ebovee09, Jesse @kierisi, Joshua @jrosenberg6432, Isabella @ivelasq3, and me @RyanEs.

See you in two weeks for our next post! Josh, with help from Ryan, Emily, Jesse, Joshua, and Isabella

Ryan A. Estrellado is a public education leader and data scientist helping administrators use practical data analysis to improve the student experience.
Emily A. Bovee, Ph.D., is an educational data scientist working in dental education.
Jesse Mostipak, M.Ed., is a community advocate, Kaggle educator, and data scientist.
Isabella C. Velásquez, MS, is a data analyst committed to nonprofit work with the aim of reducing racial and socioeconomic inequities.

Learning R With Education Datasets

Thu, 11 Jun 2020 00:00:00 +0000

Ryan A. Estrellado is a public education leader and data scientist helping administrators use practical data analysis to improve the student experience.

Timothy Gallwey wrote in The Inner Game of Tennis:

…There is a natural learning process which operates within everyone, if it is allowed to. This process is waiting to be discovered by all those who do not know of its existence … It can be discovered for yourself, if it hasn’t been already. If it has been experienced, trust it.

Discovering a new R concept like a function or package is exciting. You never know if you’re about to learn something that fundamentally changes the way you code or solve data science problems. But I get even more excited when I see somebody use new R concepts. For example, I learned about random forest models when I read about them in An Introduction to Statistical Learning (ISL). Then I imagined myself using them when I watched Julia Silge fit a random forest model to predict attendance at NFL games. I need the reading to give me language for what I see data scientists do. Then I need to see what data scientists do for me to imagine myself doing what I’ve read.

Still, for most people using R in their jobs, there’s another step. They have to imagine how to apply what they’ve read and seen to the problems they’re solving at work. But what if we used education datasets to help them imagine using R on the job, just as the authors of ISL use words and code to teach about models and Julia Silge uses video to inspire coding?

We learned from writing Data Science in Education Using R (DSIEUR) that we can combine words, code, and professional context. Professional context includes scenarios, language, and data that readers will recognize in their education jobs. We wanted readers to feel motivated and engaged by seeing words and data that reminds them of their everyday work tasks. This connection to their professional lives is a hook for readers as they engage R syntax which is, if you’ve never used it, literally a foreign language.

Let’s use pivot_longer() as an example. We’ll describe this process in three steps: discovering the concept, seeing how the concept is used, and seeing how the concept is used in education.

Step 1: See the concept

When I read something like “Use pivot_longer() to transform a dataset from wide to long”, I can imagine the shape of a dataset changing. But it’s harder to imagine what happens with the variables and their contents as the dataset’s shape changes. I’ve been using R for over five years and I still struggle to visualize the contents of many columns rearranging themselves into one.

Step 2: See how the concept is used

The concept gets much clearer when you add an example—even one with little context—to the explanation. Here’s one from the pivot_longer() vignette, which you can view with vignette("pivot"):

library(tidyverse)

# Simplest case where column names are character data
relig_income

#> # A tibble: 18 x 11
#>    religion `<$10k` `$10-20k` `$20-30k` `$30-40k` `$40-50k` `$50-75k` `$75-100k`
#>    <chr>      <dbl>     <dbl>     <dbl>     <dbl>     <dbl>     <dbl>      <dbl>
#>  1 Agnostic      27        34        60        81        76       137        122
#>  2 Atheist       12        27        37        52        35        70         73
#>  3 Buddhist      27        21        30        34        33        58         62
#>  4 Catholic     418       617       732       670       638      1116        949
#>  5 Don’t k…      15        14        15        11        10        35         21
#>  6 Evangel…     575       869      1064       982       881      1486        949
#>  7 Hindu          1         9         7         9        11        34         47
#>  8 Histori…     228       244       236       238       197       223        131
#>  9 Jehovah…      20        27        24        24        21        30         15
#> 10 Jewish        19        19        25        25        30        95         69
#> 11 Mainlin…     289       495       619       655       651      1107        939
#> 12 Mormon        29        40        48        51        56       112         85
#> 13 Muslim         6         7         9        10         9        23         16
#> 14 Orthodox      13        17        23        32        32        47         38
#> 15 Other C…       9         7        11        13        13        14         18
#> 16 Other F…      20        33        40        46        49        63         46
#> 17 Other W…       5         2         3         4         2         7          3
#> 18 Unaffil…     217       299       374       365       341       528        407
#> # … with 3 more variables: `$100-150k` <dbl>, `>150k` <dbl>, `Don't
#> #   know/refused` <dbl>

relig_income %>%
 pivot_longer(-religion, names_to = "income", values_to = "count")

#> # A tibble: 180 x 3
#>    religion income             count
#>    <chr>    <chr>              <dbl>
#>  1 Agnostic <$10k                 27
#>  2 Agnostic $10-20k               34
#>  3 Agnostic $20-30k               60
#>  4 Agnostic $30-40k               81
#>  5 Agnostic $40-50k               76
#>  6 Agnostic $50-75k              137
#>  7 Agnostic $75-100k             122
#>  8 Agnostic $100-150k            109
#>  9 Agnostic >150k                 84
#> 10 Agnostic Don't know/refused    96
#> # … with 170 more rows

Sharing an idea by pairing an abstract programming concept with a reproducible example is a common practice for experienced R programmers. Community guidelines for Stack Overflow posts and the {reprex} package are two artifacts of a popular R community norm: help folks understand an idea by using words and code.

Step 3: See how the concept is used in education

Combining the explanation with a reproducible example makes pivot_longer() more concrete by showing how it works. What happens when we connect the explanation and reproducible example to the everyday work of a data scientist in education?

In chapter seven of DSIEUR, we use pivot_longer() to transform a dataset of coursework survey responses from wide to long. Before using pivot_longer(), the dataset had a column for each survey question. When we use pivot_longer(), the name of each survey question moves to a new column called “question”. Another new column is added, “response”, which contains the corresponding response to each survey question.

To run this code, you’ll need the DSIEUR companion R package, {dataedu}:

# Install the {dataedu} package if you don't have it
# devtools::install_github("data-edu/dataedu")
library(dataedu)

Here’s the survey data in its original, wide format:

# Wide format
pre_survey

#> # A tibble: 1,102 x 12
#>    opdata_username opdata_CourseID Q1Maincellgroup… Q1Maincellgroup…
#>    <chr>           <chr>                      <dbl>            <dbl>
#>  1 _80624_1        FrScA-S116-01                  4                4
#>  2 _80623_1        BioA-S116-01                   4                4
#>  3 _82588_1        OcnA-S116-03                  NA               NA
#>  4 _80623_1        AnPhA-S116-01                  4                3
#>  5 _80624_1        AnPhA-S116-01                 NA               NA
#>  6 _80624_1        AnPhA-S116-02                  4                2
#>  7 _80624_1        AnPhA-T116-01                 NA               NA
#>  8 _80624_1        BioA-S116-01                   5                3
#>  9 _80624_1        BioA-T116-01                  NA               NA
#> 10 _80624_1        PhysA-S116-01                  4                4
#> # … with 1,092 more rows, and 8 more variables: Q1MaincellgroupRow3 <dbl>,
#> #   Q1MaincellgroupRow4 <dbl>, Q1MaincellgroupRow5 <dbl>,
#> #   Q1MaincellgroupRow6 <dbl>, Q1MaincellgroupRow7 <dbl>,
#> #   Q1MaincellgroupRow8 <dbl>, Q1MaincellgroupRow9 <dbl>,
#> #   Q1MaincellgroupRow10 <dbl>

The third through eighth columns are named after each survey question—“Q1MaincellgroupRow1”, “Q1MaincellgroupRow2”, “Q1MaincellgroupRow3”, etc. These are the column names we’ll be moving to a single column called “question” when the dataset transforms from wide to long.

Here’s the new dataset, where a column called “question” contains the question names and a column called “response” contains the corresponding responses:

# Pivot the dataset from wide to long format
pre_survey %>%
  pivot_longer(cols = Q1MaincellgroupRow1:Q1MaincellgroupRow10,
               names_to = "question",
               values_to = "response")

#> # A tibble: 11,020 x 4
#>    opdata_username opdata_CourseID question             response
#>    <chr>           <chr>           <chr>                   <dbl>
#>  1 _80624_1        FrScA-S116-01   Q1MaincellgroupRow1         4
#>  2 _80624_1        FrScA-S116-01   Q1MaincellgroupRow2         4
#>  3 _80624_1        FrScA-S116-01   Q1MaincellgroupRow3         4
#>  4 _80624_1        FrScA-S116-01   Q1MaincellgroupRow4         1
#>  5 _80624_1        FrScA-S116-01   Q1MaincellgroupRow5         5
#>  6 _80624_1        FrScA-S116-01   Q1MaincellgroupRow6         4
#>  7 _80624_1        FrScA-S116-01   Q1MaincellgroupRow7         1
#>  8 _80624_1        FrScA-S116-01   Q1MaincellgroupRow8         5
#>  9 _80624_1        FrScA-S116-01   Q1MaincellgroupRow9         5
#> 10 _80624_1        FrScA-S116-01   Q1MaincellgroupRow10        5
#> # … with 11,010 more rows

When you put it all together, the learning thought process is something like this:

There’s a function called pivot_longer(), which turns a wide dataset into a long dataset
pivot_longer() does this by putting multiple column names into its own column, then creating a new column that pairs each column name with a value
I can use pivot_longer() to change an education survey dataset that has question names for columns into one that has a “question” column and a “response” column

We’ll be back with the next post in about two weeks. Until then, do share with us about the people and tools that inspire you to work on collaborative projects. You can reach us on Twitter: Emily @ebovee09, Jesse @kierisi, Joshua @jrosenberg6432, Isabella @ivelasq3 and me @RyanEs.

Community and Collaboration: Writing Our Book in the Open

Tue, 26 May 2020 00:00:00 +0000

Ryan A. Estrellado is a public education leader and data scientist helping administrators use practical data analysis to improve the student experience.

Chicken Farm in the Open

In 2017, Emily Bovee, Jesse Mostipak, Joshua Rosenberg, Isabella Velásquez, and I started work on our book, Data Science in Education Using R (DSIEUR). We had two goals for DSIEUR. First, we aimed to write a practical reference for data scientists in education that helps them learn and apply R skills in their jobs. Second, we wanted to share the process with the R community by writing the book in the open on GitHub. After working together for almost three years, my co-authors and I submitted the manuscript for DSIEUR to Routledge and are now gearing up to begin editing the print version. The print version will be out from Routledge in late 2020, but you can read the online version of DSIEUR now at https://datascienceineducation.com/.

With the writing done, we’re reflecting on lessons we’ve learned from writing DSIEUR. In the coming weeks, we’ll share these reflections on R Views as a series of blog posts. These posts are about the people and tools in the R community that inspired us to do a book like DSIEUR. Think of these as our personal notes, typed up to help us organize our thoughts about what made this project possible. We’ll share in four parts:

Part 1: Teaching R Using Everyday Examples in Education

Learning R on the job presents many challenges, but one in particular sticks out. Once you start coding, it’s not obvious how to apply that code in everyday tasks at the office. We wrote DSIEUR to answer the question, “How would it feel to have a book that taught programming concepts, provided reproducible code, and used scenarios that data scientists in education recognize?” In this first post, we’ll explore how we put these elements together and what we learned in the process.

Part 2: How the R Community Inspired Us to Write About Data Science in Education

It wasn’t long before our team encountered our first writing challenge: do we describe our audience as “data scientists in education” or “education data scientists?”. The debate was a symbol for a larger dilemma–what common language do you use when projects like ours aren’t yet common? It helped that the community we were writing for inspired us to explore the topic. The things we love the most about the R community–welcoming folks from different backgrounds, a collective love of side projects, and a willingness to work in the open–made it safe for us to try new things and learn. We listened to stories from data scientists in education, spent a lot of time reading the Twitter #rstats hashtag, and invited community members to join the conversation. In this post, we’ll explore how community participation empowered our writing process.

Part 3: Writing In the Open

This post is about coordinating people and tools to write an open book, a challenging proposition for five writers who had only just met on Twitter. For instance, how would five people in different time zones write instructional materials and code together? And if coordinating five authors wasn’t hard enough, how would they invite the rest of the community to join the mission? Fortunately, people and programming tools encouraged us to believe that this project was possible. R, RStudio, {bookdown}, and Git had already solved publishing and collaboration problems for many. Except for some initial coding gaffes you’d expect from a team finding their feet and the occasional burned down fork, these tools freed us to focus on the larger task at hand: finding a common language for data science in education. We’ll close this post by discussing how books, authored through an open-source approach, can serve as an innovative platform for sharing knowledge with a wider audience.

Conclusion: One Writer, Five Authors.

How do you get five points of view to sound like a single voice? You’ll need a flexible sense of clarity, which I think is what Jesse meant when she said in a recent team call, “I have strong opinions, loosely held.” And it helps to have some basic rules as guardrails to flank your team as you march towards your writing deadlines. In this last post, we’ll share the workflows and processes we leaned on to discover what we wanted this book to be. We’ll also share our go-to tactics to keep the work going for the long haul, like managing meeting agendas, creating flexible norms for participation, and playing to individual strengths.

We’ll be back with that first post in about two weeks. Until then, do share with us about the people and tools that inspire you to work on collaborative projects. You can reach us on Twitter: Emily @ebovee09, Jesse @kierisi, Joshua @jrosenberg6432, Isabella @ivelasq3, and me @RyanEs.

See you in two weeks!

Ryan, with help from Emily, Jesse, Joshua, and Isabella

Emily A. Bovee, Ph.D., is an educational data scientist working in dental education.
Jesse Mostipak, M.Ed., is a community advocate, Kaggle educator and data scientist.
Joshua M. Rosenberg, Ph.D., is Assistant Professor of STEM Education and the University of Tennessee, Knoxville.
Isabella C. Velásquez, MS, is a data analyst committed to nonprofit work with the aim of reducing racial and socioeconomic inequities.