In doomlab/learnSEM: Learning Tutorials for Structural Equation Modeling

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

knitr::opts_chunk$set(echo = TRUE)

Overall Note

• You can learn R.
• You will get frustrated.
• You will get errors that don't help or make sense.
• Google is your friend.
• Try Googling the specific error message first.
• Then try googling your specific function and the error.
• Try a bunch of different search terms.

Helpful Websites

• Quick-R: www.statmethods.net
• R documentation: www.rdocumentation.org
• Swirl: www.swirlstats.com
• Stack Overflow: www.stackoverflow.com
• Learn Statistics with R: https://learningstatisticswithr.com/

Download Requirements

• Get R: http://cran.r-project.org/
• Mac: XQuartz https://www.xquartz.org/ (if you find you do not have it)
• RStudio: http://www.rstudio.org/

Outline

• Commands
• Object Types
• Subsetting
• Missing Data
• Working Directories
• Packages
• Functions

Commands

• Commands are the code that you tell R to do for you.
• They can be very simple or complex.
• Computers do what you tell them to do. Mistakes happen!
• Maybe it's a typo, maybe it's a misunderstanding of what the code does

Commands

• You can type a command directly into the console
• You can type in a document (Script or Markdown) and tell it to then run in the console

X <- 4

Commands

• > indicates the console is ready for more code
• + indicates that you haven't finished a code block
• Capitalization and symbols matter
• = and <- are equivalent
• Hit the up arrow – you can scroll through the last commands that were run
• Hit the tab key – you'll get a list of variable names and options to select from
• Use the ? followed by a command to learn more about it

Commands

• Let's take a look and run some simple commands
• Where is the console
• How do I move them around?
• How do I run code?
• What is a Script?
• What is Markdown?
• How do I run code in those?

RStudio

• What are all the windows in RStudio?
• Working Area:
• Current files that are open like scripts, markdown, etc.
• Console, Terminal, Jobs
  • Where the magic happens
  • Where everything runs
• Environment, History, ...others
• Tells you what is saved in your working environment
• What variables and types of variables you have made
• Allows you to click to view them
• Files, Plots, Packages, Help, Viewer
• Shows you a file viewer, pictures/plots, packages, and help!

Object Types

• Here are some of the basics:
• Vectors
• Lists
• Matrices
• Data Frames
• Within those objects, values can be:
• Character
• Factor (a special type of character)
• Numeric/Integer/Complex
• Logical (True, False)
• NaN (versus NA)
• Last, objects can have attributes (names)

Objects Example

library(palmerpenguins)
data(penguins)
attributes(penguins)

Objects Example

str(penguins)

names(penguins) #ls(penguins) provides this as well 

Vectors

• You can think about a vector as one row or column of data
• All the objects must be the same class
• If you try to mix and match, it will coerce them into the same type or make them NA if not.

X

• [1] indicates the number of the first item for each printed row

penguins$species

Vector Examples

A <- 1:20
A

B <- seq(from = 1, to = 20, by = 1)
B

C <- c("cheese", "is", "great")
C

D <- rep(1, times = 30)
D

Value Types

class(A)
class(C)
class(penguins)
class(penguins$species)

Functions Vary

• Functions are the commands we are running --> things like class(), rep()
• The code typed into the () is called the arguments
• Their output varies based on the type of variable you put in the arguments

dim(penguins) #rows, columns
length(penguins)
length(penguins$species)

Lists

• While vectors are one row of data, we might want to have multiple rows or types
• With a vector, it is key to understand they have to be all the same type
• Lists are a grouping of variables that can be multiple types (between list items) and can be different lengths
• Often function output is saved as a list for this reason
• They usually have names to help you print out just a small part of the list

output <- lm(flipper_length_mm ~ bill_length_mm, data = penguins)
str(output)
output$coefficients

Dimensional Data

• Matrices
• Matrices are vectors with dimensions (like a 2X3)
• All the data must be the same type
• Data Frames / Tibbles
• Like a matrix, but the columns can be different types of classes

Matrix

• Let's talk about the [ , ]
• [row, column] to subset or grab specific values

myMatrix <- matrix(data = 1:10,
                   nrow = 5,
                   ncol = 2)
myMatrix

Data Frames

• With data frames, we can use [ , ]
• However, they also have attributes that allow us to use the $ (lists have this too!)

penguins[1, 2:3]
penguins$sex[4:25] #why no comma?

Dimensional Data

• What if you want to combine data? We've already talked about c().
• rbind() allows you to put together rows
• cbind() allows you to put together columns

X <- 1:5
Y <- 6:10
# I can use either because they are the same size 
cbind(X,Y)
rbind(X,Y)

Remind R Where Things Are

• Just because you know we have penguins open and there's a variable in in called species ... you cannot just use species

ls()
ls(penguins)

Converting Object Types

• You can use as. functions to convert between types
• Show as. to see what is available
• Be careful though!

newDF <- as.data.frame(cbind(X,Y))
str(newDF)
as.numeric(c("one", "two", "3"))

Subsetting

• Subsetting is parceling out the rows/columns that you need given some criteria.
• We already talked about how to select one row/column with [1,] or [,1] and the $ operator.
• What about cases you want to select based on scores, missing data, etc.?

Subsetting Examples

• How does the logical operator work?
• It analyzes each row/column for the appropriate logical question
• We are asking when bill length is greater than 54
• We only got back the rows that the length was greater than 54
• Careful where you put it (before the ,)

penguins[1:2,] #just the first two rows 
penguins[penguins$bill_length_mm > 54 , ] #how does this work?
penguins$bill_length_mm > 54

Subsetting Examples

#you can create complex rules
penguins[penguins$bill_length_mm > 54 & penguins$bill_depth_mm > 17, ]
#you can do all BUT
penguins[ , -1]
#grab a few columns by name
vars <- c("bill_length_mm", "sex")
penguins[ , vars]

Subsetting

#another function
#notice any differences? 
subset(penguins, bill_length_mm > 54)
#other functions include filter() in tidyverse

Missing Values

• Missing values are marked with NA
• NaN stands for not a number, which doesn't automatically convert to missing
• Most functions have an option for excluding the NA values but they can be slightly different

head(complete.cases(penguins)) #creates logical
head(na.omit(penguins)) #creates actual rows
head(is.na(penguins$body_mass_g)) #for individual vectors

Working Directories

• Your computer has files and folders, and you have to tell R where to look
• The working directory is where you are currently telling it to look

getwd()

Working Directory

• You can set the working directory by doing something like this
• I would suggest this is pretty error prone and breaks when you move files!

setwd("/Users/buchanan/OneDrive - Harrisburg University/Teaching/ANLY 580/updated/1 Introduction R")

Working Directory

• Working directories are critical because they allow you to automate
• Instead of using the point and click options, you can just run code to open your specific files
• Markdown files are the best!
• Projects are the best!

Importing Files

• There are many ways to import files
• You can use base R functions (readLines, read.csv)
• You can use tidyverse (read_csv)
• You can use Import Dataset clickable option
• Why not use one package that does most of it like magic?

library(rio)
myDF <- import("data/assignment_introR.csv")
head(myDF)

Packages

• You can install extra functions by installing packages or libraries
• These can be downloaded from CRAN using install.packages()
• You can also install these by using the Packages tab
• Additional packages can be installed from GitHub and other places

install.packages("car")

Packages

• View what is installed with the Packages window
• Every time you get a major R update, you will likely have to reinstall packages
• Every time you restart R, you will need to reload each package
• Helpful to put the library code right at the top of your scripts

library(car)

Functions

• Functions are pre-written code to help you run analyses
• Get help with function, learn what the arguments should be:
• Let's flip back to RStudio to see what this did

?lm
help(lm)

Functions

args(lm)
example(lm)

Define Your Own Function

• Name the function before <-
• Define the arguments inside ()
• Define what the function does inside {}

pizza <- function(x){ x^2 }
pizza(3)

Example Functions

table(penguins$species)
summary(penguins$bill_length_mm)

Examples Functions with Missing Data

mean(penguins$bill_length_mm) #returns NA
mean(penguins$bill_length_mm, na.rm = TRUE)

cor(penguins[ , c("bill_length_mm", "bill_depth_mm", "flipper_length_mm")])
cor(penguins[ , c("bill_length_mm", "bill_depth_mm", "flipper_length_mm")],
    use = "pairwise.complete.obs")

Other Descriptive Functions

• cov()
• var()
• sd()
• scale()

Wrapping Up

• In this demo, you've learned:
• Some basic programming terminology
• Specific R defaults and issues
• Example functions and use cases
• How do I get started?
• Practice!

doomlab/learnSEM documentation built on Jan. 25, 2024, 2 p.m.