• 4. Data in R summary

Links to: Summary, Chatbot Tutor, Practice Questions, Glossary, R functions, R packages introduced, and Additional resources.

Chapter summary

A beautiful Clarkia xantiana flower.

Keeping data in the tidy format—where each column represents a variable and each row represents an observation—allows you to fully leverage the powerful tools of the tidyverse. In the tidyverse, data are stored in tibbles, a modern update to data frames that enhances readability and maintains consistent data types. The dplyr package offers a suite of intuitive functions for transforming and analyzing data. These functions include: mutate() for adding or modifying columns.

• select() for choosing columns.
  
• filter() for subsetting rows, and
  
• rename(). for changing column names.

Together – especially when used with the pipe operator—these – tools enable clear, reproducible workflows. In the next two chapters we will begin summarizing data with dplyr tools.

Chatbot tutor

Please interact with this custom chatbot (ChatGPT link here, Gemini link here) I have made to help you with this chapter. I suggest interacting with at least ten back-and-forths to ramp up and then stopping when you feel like you got what you needed from it.

Practice Questions

Try these questions!

Q1) Consider the table below. The data are tidynot tidy

location	GC	GC	GC	GC	GC	GC
ril	A1	A100	A102	A104	A106	A107
mean_visits	0.0000	0.1875	0.2500	0.0000	0.0000	0.0000

Here the data are transposed, so the data are not tidy. Remember in tidy data each variable is a column, not a row. This is particularly hard for R because there are numerous types of data in a column.

.

---

Q2) Consider the table below. The data are tidynot tidy

location-ril	mean_visits
GC-A1	0.0000
GC-A100	0.1875
GC-A102	0.2500
GC-A104	0.0000
GC-A106	0.0000
GC-A107	0.0000

Here location and ril are combined in a single column, so the data are not tidy. Remember in tidy data each variable is its own column. It would be hard to get e.g. means for RILs of locations in this format.

---

Q3) Consider the table below. The data are tidynot tidy

ril	GC	SR
A1	0.0000	0.6667
A100	0.1875	0.5833
A102	0.2500	0.6667
A104	0.0000	1.7500
A106	0.0000	0.5000
A107	0.0000	1.5000

This is known as “wide format” and is not tidy. Here the variable, location, is used as a column heading. This can be a fine way to present data to people, but it’s not how we are analyzing data.

---

Q4) You should always make sure data are tidy when (pick best answer)

collecting data presenting data analyzing data with dplyr all of the above

---

For the following questions consider the iris data set built into R. Preview below.

Sepal.Length	Sepal.Width	Petal.Length	Petal.Width	Species
5.1	3.5	1.4	0.2	setosa
4.9	3.0	1.4	0.2	setosa
4.7	3.2	1.3	0.2	setosa
4.6	3.1	1.5	0.2	setosa

Q5) Consider the table above and code below. What is the value of pl_pw_ratio in the first row?

iris |> 
  mutate(pl_pw_ratio = Petal.Length / Petal.Width)

---

Q6) Consider the code below. Which column names will appear in the output? (select all that apply).

Sepal.Length Sepal.Width Petal.Length Petal.Width Species pl_pw_ratio

iris |> 
  mutate(pl_pw_ratio = Petal.Length / Petal.Width) |>
  select(Species, pl_pw_ratio)

---

Q7) Say I wanted to use clean_names() from janitor. Which two of the scripts below would work? A and BA and CA and DB and CB and DC and D

# A
iris |> 
  clean_names() |>
  mutate(pl_pw_ratio = Petal.Length / Petal.Width)

# B
iris |> 
  mutate(pl_pw_ratio = Petal.Length / Petal.Width)|>
  clean_names() 

# C
iris |> 
  clean_names()  |>
  mutate(pl_pw_ratio = petal_length / petal_width)

# D
iris |> 
  clean_names()  |>
  mutate(pl_pw_ratio = Petal.Length / Petal.Width)

Use the R environment below to answer the next set of questions

As you can see, the code above returns the error: "Error: object ‘Species’ not found."

Q8) What was this code aiming to do?

Create a new column called Species Convert Species to a factor Return only the rows of the iris dataset where Species is “setosa” Remove the Species column from the dataset Sort the dataset by Species

---

Q9) Why did the error, "Error: object ‘Species’ not found." arise?

We use select() to select by values in a row, not filter() The iris dataset does not contain a Species column The Species column was deleted in the setup chunk filter() only works on tibbles, not data frames Species must be quoted inside filter() R used stats::filter() instead of dplyr::filter(), so Species was looked for as an object (not a column)

Try typing glimpse(iris)

---

Q10) How to fix the error "Error: object ‘Species’ not found."? Find both correct answers `

Convert iris to a tibble first:
as_tibble(iris) |> filter(Species == “setosa”)

Tell R to use dplyr's filter:
iris|>dplyr::filter(Species == “setosa”)

Rename species to Species:
iris|>rename(Species=species)|>filter(Species==“setosa”)

Filter on species, not Species:
iris|>filter(species == “setosa”)

Tell R to use dplyr's filter. After library(dplyr), type:
conflict_prefer(“filter”,“dplyr”)

Debugger’s note: In a hidden setup chunk I wrote: conflict_prefer("filter", "stats") to tell R to prefer stats::filter() over dplyr::filter(). Because of this, filter() no longer understands column names inside a data frame. So I forced this error on you.

In real workflows, it can be hard to predict which version of a function R will use. This issue arises even when I don’t trick you. So always tell R which version of filter() you want to use. filter() may on your computer today without dplyr::filter() or conflict_prefer("filter","dplyr"), but it might not work on someone else computer (it might even fail on your computer next week).

Glossary of Terms

• Tidy Data A structured format where:
  • Each row represents an observation.
    
  • Each column represents a variable.
    
  • Each cell contains a single measurement.
• Tibbles: A modern form of a data frame in R with:
  • Cleaner printing (only first 10 rows, fits columns to screen).
    
  • Explicit display of data types (e.g., , ).
    
  • Strict subsetting (prevents automatic type conversion).
    
  • Character data is not automatically converted to factors.
• Piping (|>) functions: A way to chain operations together, making code more readable and modular.
  
• Missing Data (NA): R uses NA to represent missing values. Operations with NA return NA unless handled explicitly (e.g., na.rm = TRUE to ignore missing values, use = "pairwise.complete.obs", etc).
  
• Warnings: Indicate a possible issue but allow code to run (e.g., NAs introduced by coercion).
  
• Errors: Stop execution completely when something is invalid.

---

Key R functions

• read_csv() (readr): Reads a CSV file into R as a tibble, automatically guessing column types.

• select() (dplyr): Selects specific columns from a dataset.

• mutate() (dplyr): Creates or modifies columns in a dataset.

• dplyr::filter() (dplyr. Because there is a different filter in a different package, always specify which filter you want r to use.): Only retains rows wihich meet some logical criteria.

• rename()` (dplyr: Allows you to manually changes column names.

• as.numeric(): Converts a vector to a numeric data type.

• |> (Base R Pipe Operator): Passes the result of one function into another, making code more readable.

---

R Packages Introduced

• readr: A tidyverse package for reading rectangular data files (e.g., read_csv()).

• dplyr: A tidyverse package for data manipulation, including mutate(), glimpse(), and across().

Additional resources

R Recipes:

• Selecting columns.
  
• Add a new column (or modify an existing one).

Other web resources:

• Chapter 10: Tidy data from R for data science (Grolemund & Wickham (2018)).
  
• Animated dplyr functions from R or the rest of us.

Videos:

• Basic Data Manipulation (From Stat454).

• Calculations on tibble (From Stat454).

Grolemund, G., & Wickham, H. (2018). R for data science.