<p>One of the best ways to improve your reach as a data scientist is to write functions. Functions allow you to automate common tasks in a more powerful and general way than copy-and-pasting. Writing a function has three big advantages over using copy-and-paste:</p>
<oltype="1"><li><p>You can give a function an evocative name that makes your code easier to understand.</p></li>
<li><p>As requirements change, you only need to update code in one place, instead of many.</p></li>
<li><p>You eliminate the chance of making incidental mistakes when you copy and paste (i.e.updating a variable name in one place, but not in another).</p></li>
</ol><p>A good rule of thumb is to consider writing a function whenever you’ve copied and pasted a block of code more than twice (i.e.you now have three copies of the same code). In this chapter, you’ll learn about three useful types of functions:</p>
<ul><li>Vector functions take one or more vectors as input and return a vector as output.</li>
<li>Data frame functions take a data frame as input and return a data frame as output.</li>
<li>Plot functions that take a data frame as input and return a plot as output.</li>
</ul><p>Each of these sections include many examples to help you generalize the patterns that you see. These examples wouldn’t be possible without the help of folks of twitter, and we encourage follow the links in the comment to see original inspirations. You might also want to read the original motivating tweets for <ahref="https://twitter.com/hadleywickham/status/1571603361350164486">general functions</a> and <ahref="https://twitter.com/hadleywickham/status/1574373127349575680">plotting functions</a> to see even more functions.</p>
<p>We’ll wrap up a variety of functions from around the tidyverse. We’ll also use nycflights13 as a source of familiar data to use our functions with.</p>
<p>We’ll begin with vector functions: functions that take one or more vectors and return a vector result. For example, take a look at this code. What does it do?</p>
<p>You might be able to puzzle out that this rescales each column to have a range from 0 to 1. But did you spot the mistake? When Hadley wrote this code he made an error when copying-and-pasting and forgot to change an <code>a</code> to a <code>b</code>. Preventing this type of mistake of is one very good reason to learn how to write functions.</p>
<p>To write a function you need to first analyse your repeated code to figure what parts are constant and what parts vary. If we take the code above and pull it outside of <code><ahref="https://dplyr.tidyverse.org/reference/mutate.html">mutate()</a></code>, it’s a little easier to see the pattern because each repetition is now one line:</p>
<p>To turn this into a function you need three things:</p>
<oltype="1"><li><p>A <strong>name</strong>. Here we’ll use <code>rescale01</code> because this function rescales a vector to lie between 0 and 1.</p></li>
<li><p>The <strong>arguments</strong>. The arguments are things that vary across calls and our analysis above tells us that we have just one. We’ll call it <code>x</code> because this is the conventional name for a numeric vector.</p></li>
<li><p>The <strong>body</strong>. The body is the code that’s repeated across all the calls.</p></li>
<p>(In <ahref="#chp-iteration"data-type="xref">#chp-iteration</a>, you’ll learn how to use <code><ahref="https://dplyr.tidyverse.org/reference/across.html">across()</a></code> to reduce the duplication even further so all you need is <code>df |> mutate(across(a:d, rescale01))</code>).</p>
<p>You might notice that the <code>rescale01()</code> function does some unnecessary work — instead of computing <code><ahref="https://rdrr.io/r/base/Extremes.html">min()</a></code> twice and <code><ahref="https://rdrr.io/r/base/Extremes.html">max()</a></code> once we could instead compute both the minimum and maximum in one step with <code><ahref="https://rdrr.io/r/base/range.html">range()</a></code>:</p>
<p>That result is not particularly useful so we could ask <code><ahref="https://rdrr.io/r/base/range.html">range()</a></code> to ignore infinite values:</p>
<p>These changes illustrate an important benefit of functions: because we’ve moved the repeated code into a function, we only need to make the change in one place.</p>
<p>Now you’ve got the basic idea of functions, let’s take a look at a whole bunch of examples. We’ll start by looking at “mutate” functions, i.e.functions that work well inside of <code><ahref="https://dplyr.tidyverse.org/reference/mutate.html">mutate()</a></code> and <code><ahref="https://dplyr.tidyverse.org/reference/filter.html">filter()</a></code> because they return an output of the same length as the input.</p>
<p>Let’s start with a simple variation of <code>rescale01()</code>. Maybe you want to compute the Z-score, rescaling a vector to have a mean of zero and a standard deviation of one:</p>
<p>Or maybe you want to wrap up a straightforward <code><ahref="https://dplyr.tidyverse.org/reference/case_when.html">case_when()</a></code> and give it a useful name. For example, this <code>clamp()</code> function ensures all values of a vector lie in between a minimum or a maximum:</p>
<p>Of course functions don’t just need to work with numeric variables. You might want to do some repeated string manipulation. Maybe you need to make the first character upper case:</p>
<p>Sometimes your functions will be highly specialized for one data analysis step. For example, if you have a bunch of variables that record missing values as 997, 998, or 999, you might want to write a function to replace them with <code>NA</code>:</p>
<p>We’ve focused on examples that take a single vector because we think they’re the most common. But there’s no reason that your function can’t take multiple vector inputs. For example, you might want to compute the distance between two locations on the globe using the haversine formula. This requires four vectors:</p>
<p>Another important family of vector functions is summary functions, functions that return a single value for use in <code><ahref="https://dplyr.tidyverse.org/reference/summarise.html">summarize()</a></code>. Sometimes this can just be a matter of setting a default argument or two:</p>
<p>You can also write functions with multiple vector inputs. For example, maybe you want to compute the mean absolute prediction error to help you compare model predictions with actual values:</p>
</h1><p>Once you start writing functions, there are two RStudio shortcuts that are super useful:</p><ul><li><p>To find the definition of a function that you’ve written, place the cursor on the name of the function and press <code>F2</code>.</p></li>
<li><p>To quickly jump to a function, press <code>Ctrl + .</code> to open the fuzzy file and function finder and type the first few letters of your function name. You can also navigate to files, Quarto sections, and more, making it a very handy navigation tool.</p></li>
</ul></div>
</section>
<sectionid="exercises"data-type="sect2">
<h2>
Exercises</h2>
<oltype="1"><li>
<p>Practice turning the following code snippets into functions. Think about what each function does. What would you call it? How many arguments does it need?</p>
<li><p>In the second variant of <code>rescale01()</code>, infinite values are left unchanged. Can you rewrite <code>rescale01()</code> so that <code>-Inf</code> is mapped to 0, and <code>Inf</code> is mapped to 1?</p></li>
<li><p>Given a vector of birthdates, write a function to compute the age in years.</p></li>
<li><p>Write your own functions to compute the variance and skewness of a numeric vector. Variance is defined as <spanclass="math display">\[
\]</span> where <spanclass="math inline">\(\bar{x} = (\sum_i^n x_i) / n\)</span> is the sample mean. Skewness is defined as <spanclass="math display">\[
<li><p>Write <code>both_na()</code>, a summary function that takes two vectors of the same length and returns the number of positions that have an <code>NA</code> in both vectors.</p></li>
<li>
<p>Read the documentation to figure out what the following functions do. Why are they useful even though they are so short?</p>
<p>Vector functions are useful for pulling out code that’s repeated within a dplyr verb. But you’ll often also repeat the verbs themselves, particularly within a large pipeline. When you notice yourself copying and pasting multiple verbs multiple times, you might think about writing a data frame function. Data frame functions work like dplyr verbs: they take a data frame as the first argument, some extra arguments that say what to do with it, and return a data frame or vector.</p>
<p>To let you write a function that uses dplyr verbs, we’ll first introduce you to the challenge of indirection and how you can overcome it with embracing, <code>{{ }}</code>. With this theory under your belt, we’ll then show you a bunch of examples to illustrate what you might do with it.</p>
<p>When you start writing functions that use dplyr verbs you rapidly hit the problem of indirection. Let’s illustrate the problem with a very simple function: <code>grouped_mean()</code>. The goal of this function is compute the mean of <code>mean_var</code> grouped by <code>group_var</code>:</p>
<p>Regardless of how we call <code>grouped_mean()</code> it always does <code>df |> group_by(group_var) |> summarize(mean(mean_var))</code>, instead of <code>df |> group_by(group) |> summarize(mean(x))</code> or <code>df |> group_by(group) |> summarize(mean(y))</code>. This is a problem of indirection, and it arises because dplyr uses <strong>tidy evaluation</strong> to allow you to refer to the names of variables inside your data frame without any special treatment.</p>
<p>Tidy evaluation is great 95% of the time because it makes your data analyses very concise as you never have to say which data frame a variable comes from; it’s obvious from the context. The downside of tidy evaluation comes when we want to wrap up repeated tidyverse code into a function. Here we need some way to tell <code>group_mean()</code> and <code><ahref="https://dplyr.tidyverse.org/reference/summarise.html">summarize()</a></code> not to treat <code>group_var</code> and <code>mean_var</code> as the name of the variables, but instead look inside them for the variable we actually want to use.</p>
<p>Tidy evaluation includes a solution to this problem called <strong>embracing</strong> 🤗. Embracing a variable means to wrap it in braces so (e.g.) <code>var</code> becomes <code>{{ var }}</code>. Embracing a variable tells dplyr to use the value stored inside the argument, not the argument as the literal variable name. One way to remember what’s happening is to think of <code>{{ }}</code> as looking down a tunnel — <code>{{ var }}</code> will make a dplyr function look inside of <code>var</code> rather than looking for a variable called <code>var</code>.</p>
<p>So the key challenge in writing data frame functions is figuring out which arguments need to be embraced. Fortunately, this is easy because you can look it up from the documentation 😄. There are two terms to look for in the docs which correspond to the two most common sub-types of tidy evaluation:</p>
<ul><li><p><strong>Data-masking</strong>: this is used in functions like <code><ahref="https://dplyr.tidyverse.org/reference/arrange.html">arrange()</a></code>, <code><ahref="https://dplyr.tidyverse.org/reference/filter.html">filter()</a></code>, and <code><ahref="https://dplyr.tidyverse.org/reference/summarise.html">summarize()</a></code> that compute with variables.</p></li>
<li><p><strong>Tidy-selection</strong>: this is used for functions like <code><ahref="https://dplyr.tidyverse.org/reference/select.html">select()</a></code>, <code><ahref="https://dplyr.tidyverse.org/reference/relocate.html">relocate()</a></code>, and <code><ahref="https://dplyr.tidyverse.org/reference/rename.html">rename()</a></code> that select variables.</p></li>
</ul><p>Your intuition about which arguments use tidy evaluation should be good for many common functions — just think about whether you can compute (e.g.<code>x + 1</code>) or select (e.g.<code>a:x</code>).</p>
<p>If you commonly perform the same set of summaries when doing initial data exploration, you might consider wrapping them up in a helper function:</p>
<p>(Whenever you wrap <code><ahref="https://dplyr.tidyverse.org/reference/summarise.html">summarize()</a></code> in a helper, we think it’s good practice to set <code>.groups = "drop"</code> to both avoid the message and leave the data in an ungrouped state.)</p>
<p>The nice thing about this function is, because it wraps <code><ahref="https://dplyr.tidyverse.org/reference/summarise.html">summarize()</a></code>, you can use it on grouped data:</p>
<p>Furthermore, since the arguments to summarize are data-masking also means that the <code>var</code> argument to <code>summary6()</code> is data-masking. That means you can also summarize computed variables:</p>
<p>To summarize multiple variables, you’ll need to wait until <ahref="#sec-across"data-type="xref">#sec-across</a>, where you’ll learn how to use <code><ahref="https://dplyr.tidyverse.org/reference/across.html">across()</a></code>.</p>
<p>Another popular <code><ahref="https://dplyr.tidyverse.org/reference/summarise.html">summarize()</a></code> helper function is a version of <code><ahref="https://dplyr.tidyverse.org/reference/count.html">count()</a></code> that also computes proportions:</p>
<p>This function has three arguments: <code>df</code>, <code>var</code>, and <code>sort</code>, and only <code>var</code> needs to be embraced because it’s passed to <code><ahref="https://dplyr.tidyverse.org/reference/count.html">count()</a></code> which uses data-masking for all variables in <code>…</code>.</p>
<p>Or maybe you want to find the sorted unique values of a variable for a subset of the data. Rather than supplying a variable and a value to do the filtering, we’ll allow the user to supply a condition:</p>
<p>Here we embrace <code>condition</code> because it’s passed to <code><ahref="https://dplyr.tidyverse.org/reference/filter.html">filter()</a></code> and <code>var</code> because it’s passed to <code><ahref="https://dplyr.tidyverse.org/reference/distinct.html">distinct()</a></code> and <code><ahref="https://dplyr.tidyverse.org/reference/arrange.html">arrange()</a></code>.</p>
<p>We’ve made all these examples to take a data frame as the first argument, but if you’re working repeatedly with the same data, it can make sense to hardcode it. For example, the following function always works with the flights dataset and always selects <code>time_hour</code>, <code>carrier</code>, and <code>flight</code> since they form the compound primary key that allows you to identify a row.</p>
<p>Sometimes you want to select variables inside a function that uses data-masking. For example, imagine you want to write a <code>count_missing()</code> that counts the number of missing observations in rows. You might try writing something like:</p>
<p>This doesn’t work because <code><ahref="https://dplyr.tidyverse.org/reference/group_by.html">group_by()</a></code> uses data-masking, not tidy-selection. We can work around that problem by using the handy <code><ahref="https://dplyr.tidyverse.org/reference/pick.html">pick()</a></code> function, which allows you to use tidy-selection inside data-masking functions:</p>
<p>Another convenient use of <code><ahref="https://dplyr.tidyverse.org/reference/pick.html">pick()</a></code> is to make a 2d table of counts. Here we count using all the variables in the <code>rows</code> and <code>columns</code>, then use <code><ahref="https://tidyr.tidyverse.org/reference/pivot_wider.html">pivot_wider()</a></code> to rearrange the counts into a grid:</p>
<p>While our examples have mostly focused on dplyr, tidy evaluation also underpins tidyr, and if you look at the <code><ahref="https://tidyr.tidyverse.org/reference/pivot_wider.html">pivot_wider()</a></code> docs you can see that <code>names_from</code> uses tidy-selection.</p>
<p>Converts the user supplied variable that uses clock time (e.g.<code>dep_time</code>, <code>arr_time</code>, etc.) into a decimal time (i.e.hours + (minutes / 60)).</p>
<li><p>For each of the following functions list all arguments that use tidy evaluation and describe whether they use data-masking or tidy-selection: <code><ahref="https://dplyr.tidyverse.org/reference/distinct.html">distinct()</a></code>, <code><ahref="https://dplyr.tidyverse.org/reference/count.html">count()</a></code>, <code><ahref="https://dplyr.tidyverse.org/reference/group_by.html">group_by()</a></code>, <code><ahref="https://dplyr.tidyverse.org/reference/rename.html">rename_with()</a></code>, <code><ahref="https://dplyr.tidyverse.org/reference/slice.html">slice_min()</a></code>, <code><ahref="https://dplyr.tidyverse.org/reference/slice.html">slice_sample()</a></code>.</p></li>
<p>Instead of returning a data frame, you might want to return a plot. Fortunately, you can use the same techniques with ggplot2, because <code><ahref="https://ggplot2.tidyverse.org/reference/aes.html">aes()</a></code> is a data-masking function. For example, imagine that you’re making a lot of histograms:</p>
<p>Wouldn’t it be nice if you could wrap this up into a histogram function? This is easy as pie once you know that <code><ahref="https://ggplot2.tidyverse.org/reference/aes.html">aes()</a></code> is a data-masking function and you need to embrace:</p>
<p>Note that <code>histogram()</code> returns a ggplot2 plot, meaning you can still add on additional components if you want. Just remember to switch from <code>|></code> to <code>+</code>:</p>
<p>It’s straightforward to add more variables to the mix. For example, maybe you want an easy way to eyeball whether or not a data set is linear by overlaying a smooth line and a straight line:</p>
<p>Some of the most useful helpers combine a dash of dplyr with ggplot2. For example, if you might want to do a vertical bar chart where you automatically sort the bars in frequency order using <code><ahref="https://forcats.tidyverse.org/reference/fct_inorder.html">fct_infreq()</a></code>. Since the bar chart is vertical, we also need to reverse the usual order to get the highest values at the top:</p>
<p>We have to use a new operator here, <code>:=</code>, because we are generating the variable name based on user-supplied data. Variable names go on the left hand side of <code>=</code>, but R’s syntax doesn’t allow anything to the left of <code>=</code> except for a single literal name. To work around this problem, we use the special operator <code>:=</code> which tidy evaluation treats in exactly the same way as <code>=</code>.</p>
<p>Or maybe you want to make it easy to draw a bar plot just for a subset of the data:</p>
<p>You can also get creative and display data summaries in other ways. For example, this code uses the axis labels to display the highest value. As you learn more about ggplot2, the power of your functions will continue to increase.</p>
<p>Unfortunately, programming with faceting is a special challenge, because faceting was implemented before we understood what tidy evaluation was and how it should work. So you have to learn a new syntax. When programming with facets, instead of writing <code>~ x</code>, you need to write <code>vars(x)</code> and instead of <code>~ x + y</code> you need to write <code>vars(x, y)</code>. The only advantage of this syntax is that <code><ahref="https://ggplot2.tidyverse.org/reference/vars.html">vars()</a></code> uses tidy evaluation so you can embrace within it:</p>
<p>As with data frame functions, it can be useful to make your plotting functions tightly coupled to a specific dataset, or even a specific variable. For example, the following function makes it particularly easy to interactively explore the conditional distribution of <code>carat</code> from the diamonds dataset.</p>
<p>Wouldn’t it be nice if we could label the output with the variable and the bin width that was used? To do so, we’re going to have to go under the covers of tidy evaluation and use a function from the package we haven’t talked about yet: rlang. rlang is a low-level package that’s used by just about every other package in the tidyverse because it implements tidy evaluation (as well as many other useful tools).</p>
<p>To solve the labeling problem we can use <code><ahref="https://rlang.r-lib.org/reference/englue.html">rlang::englue()</a></code>. This works similarly to <code><ahref="https://stringr.tidyverse.org/reference/str_glue.html">str_glue()</a></code>, so any value wrapped in <code><ahref="https://rdrr.io/r/base/Paren.html">{ }</a></code> will be inserted into the string. But it also understands <code>{{ }}</code>, which automatically inserts the appropriate variable name:</p>
<oltype="1"><li><p>Draw a scatterplot given dataset and <code>x</code> and <code>y</code> variables.</p></li>
<li><p>Add a line of best fit (i.e.a linear model with no standard errors).</p></li>
<li><p>Add a title.</p></li>
</ol></section>
</section>
<sectionid="style"data-type="sect1">
<h1>
Style</h1>
<p>R doesn’t care what your function or arguments are called but the names make a big difference for humans. Ideally, the name of your function will be short, but clearly evoke what the function does. That’s hard! But it’s better to be clear than short, as RStudio’s autocomplete makes it easy to type long names.</p>
<p>Generally, function names should be verbs, and arguments should be nouns. There are some exceptions: nouns are ok if the function computes a very well known noun (i.e.<code><ahref="https://rdrr.io/r/base/mean.html">mean()</a></code> is better than <code>compute_mean()</code>), or accessing some property of an object (i.e.<code><ahref="https://rdrr.io/r/stats/coef.html">coef()</a></code> is better than <code>get_coefficients()</code>). Use your best judgement and don’t be afraid to rename a function if you figure out a better name later.</p>
<p>R also doesn’t care about how you use white space in your functions but future readers will. Continue to follow the rules from <ahref="#chp-workflow-style"data-type="xref">#chp-workflow-style</a>. Additionally, <code>function()</code> should always be followed by squiggly brackets (<code><ahref="https://rdrr.io/r/base/Paren.html">{}</a></code>), and the contents should be indented by an additional two spaces. This makes it easier to see the hierarchy in your code by skimming the left-hand margin.</p>
<li><p>Make a case for why <code>norm_r()</code>, <code>norm_d()</code> etc. would be better than <code><ahref="https://rdrr.io/r/stats/Normal.html">rnorm()</a></code>, <code><ahref="https://rdrr.io/r/stats/Normal.html">dnorm()</a></code>. Make a case for the opposite.</p></li>
<p>In this chapter, you learned how to write functions for three useful scenarios: creating a vector, creating a data frames, or creating a plot. Along the way you saw many examples, which hopefully started to get your creative juices flowing, and gave you some ideas for where functions might help your analysis code.</p>
<ul><li>To learn more about programming with tidy evaluation, see useful recipes in <ahref="https://dplyr.tidyverse.org/articles/programming.html">programming with dplyr</a> and <ahref="https://tidyr.tidyverse.org/articles/programming.html">programming with tidyr</a> and learn more about the theory in <ahref="https://rlang.r-lib.org/reference/topic-data-mask.html">What is data-masking and why do I need {{?</a>.</li>
<li>To learn more about reducing duplication in your ggplot2 code, read the <ahref="https://ggplot2-book.org/programming.html"class="uri">Programming with ggplot2</a> chapter of the ggplot2 book.</li>
<li>For more advice on function style, see the <ahref="https://style.tidyverse.org/functions.html"class="uri">tidyverse style guide</a>.</li>
</ul><p>In the next chapter, we’ll dive into some of the details of R’s vector data structures that we’ve omitted so far. These are not immediately useful by themselves, but are a necessary foundation for the following chapter on iteration which gives you further tools for reducing code duplication.</p>
