General Formatting

First, I will talk about some of the general formatting “rules” in R. They are not hard and strict rules. Your R code will likely still work without following them. However, it will be extremely hard for people to understand your code.

A few typical guidelines include using <- to define objects in R and the = to clarify function arguments. Also, it is encouraged to comment and annotate your code. The way to comment in R is with the #. I typically make large breaks with section headers for variable definitions, calculations, results, and plots.

The main guideline I will stress that you follow is include white space in your code! It might not be required to make the code run, but line breaks, tabs, and spaces will always make your code so much easier to follow. As an example, let’s look at our slope function we made here.

slope<-function(X,Y,B){print(paste("The slope is equal to ",(Y-B)/X))}

We know what this function does since we have seen it before. But if the above code was how it was first shown to you, it would be more of a challenge to figure out exactly how this function works. Another example using a for loop:

# Example A
count=0
for(i in 1:10){if(i%%2==0)count=count+1}
print(count)

# Example B
count <- 0

for(i in 1:10){
  if(i%%2==0){
    count=count+1
  }
}

print(count)

What exactly are these for loops doing? Both are counting how many numbers in a collection are divisible by 2. Did it take you longer to figure out Example A because it is one big block of text? It’s very hard to read and I actually missed the if statement the first time I read through Example A. Adding in spacing is like dividing your code into paragraphs and sections.

The Tidyverse

The tidyverse is a collection of R packages that are designed for data science, according to the website moderated by Hadley Wickham. The collection of packages all aim to unify and improve the grammar of coding. You are able to install the entire tidyverse suite by running install.packages("tidyverse"). It includes eight core packages that are all wonderful in their own right. I invite you to check them all out here.

However, the one that perhaps does the most work in the tidyverse is dplyr. This package provides all of the new grammar rules for data manipulation to use within R. It keeps the general formatting rules of base R but changes the basic formatting through the use of pipes: %>%. These essentially take an object on the left side of the pipe and applies the function on the right side. In words, you take Object X (usually a data table) and apply a cleaning function. It sort of reverses the way you think about how functions work in R. But the nice thing is that you can create a list of piped functions that will be applied in sequence. This is easier to read than a series of nested functions when it comes to sharing your code.

library(dplyr)
dat <- mtcars %>% 
  as_tibble() #the data must be tidy

dat %>% 
  select(mpg, wt, hp) %>%
  filter(mpg <= 15)

Base R vs Tidy R

Similar to the Python vs. R debate, there is a debate among the R community whether or not to use the tidyverse packages. Just as before, I ride the fence on this debate acknowledging that each formatting methods has advantages and disadvantages depending on the situation.

The main issue with the general formatting guidelines in base R is that they are guidelines. There is nothing requiring or encouraging using those rules. The tidyverse encourages you to improve the formatting or grammar of your coding. When a code is created in the tidyverse, it is very consistent and readable.

However, the tidyverse is mainly focused on data manipulation. As such most of the grammatical techniques can only be used for data structures. Tidy data structures only. You need to force your data to fit into a tidy format to use any of the functions provided by the tidyverse packages. In most cases it is simple to do as can be seen in my example. However, there are some situations such as when the data sets are small and simple where using tidyverse functions actually do make things more complicated when they do not need to be. Actually, the author of this blog post puts it best (and also relates it to my main issue with Python): Using the tidyverse changes the structure of the coding language. Doing that translation from base R takes time and effort. Some feel that new packages should build on the base language not change it entirely.

In my opinion, base R formatting and functions for data manipulations should be used when learning R. The logic moves in a more straightforward way than it does with tidyverse. But if you are trying to do manipulate large, complex data sets, then the tidyverse will outline the exact steps taken to manipulate data. I love it for this exact reason. It is wonderful for replicating data analyses and remembering what steps you took and what order you took them in.