R Advanced

Note

Welcome to the Advanced R Tutorial webpage! This tutorial will cover some basic data manipulation and analysis techiniques that you will be using for this program and future jobs. These includes: export and import dataframe, libraries, and Exploratory Data Analysis(EDA).

1 Import and Export Data Set

1.1 Import Dataframe

You can either download a dataset to your computer and open it via a file path, or load it directly from an online link. For the purpose of our demonstration, we will download and load the dataset from an online resource.

Data_BigMac <- read.csv("https://calmcode.io/static/data/bigmac.csv")
TipTip

In your first semester of statistics class at GW, you will use read.csv() regularly to load in data files to R. Don’t forget it!

1.2 Export Dataframe

There are several output file types you can choose from when exporting a dataframe. We will export the dataframe into a .csv file in this example.

When exporting a dataframe, you will need to specify the file path on your computer so that R knows where to store the exported file. If you wish to follow along through the next example:

" my_documents_path <- file.path(Sys.getenv("USERPROFILE"), "Documents", "Data_BigMac.csv") "
" write.csv(Data_BigMac, my_documents_path, row.names = FALSE) "

the exported .csv file should be saved in this location: in your computer:

C:\Users\"insert your username here"\Documents

2 Libraries

Packages and libraries are additional functions or features developed by third-parties to make data analysis in R more efficient and support clearer visual interpretation of the data.

2.1 Install package and enable libraries

The syntax for installing a package is: install.packages(“the package’s name”)

# install.packages("dplyr",repos = "http://cran.us.r-project.org")
#install.packages("ggplot2",repos = "http://cran.us.r-project.org")

The syntax for enabling a library is: library(“the package’s name”)

suppressPackageStartupMessages(library("dplyr"))
suppressPackageStartupMessages(library("ggplot2"))
TipTip

We only need to install a package once and you do not need to do it again. It is a good practice to comment out the line of code where you install the package after the first time, especially if you are working on a project that you need to compile several times.

2.2 Other helpful packages

Predominant packages for R include: tidyverse, dplyr, ggplot2, psycho, and mice. We will use some of them in the next section, but first let’s install and enable them.

#install.packages("tidyverse",repos = "http://cran.us.r-project.org")
#install.packages("psycho",repos = "http://cran.us.r-project.org")
#install.packages("mice",repos = "http://cran.us.r-project.org")
#suppressPackageStartupMessages(library("tidyverse"))
#suppressPackageStartupMessages(library("psycho"))
#suppressPackageStartupMessages(library("mice"))

3 Exploratory Data Analysis

Exploratory Data Analysis (EDA) is an approach focused on understanding the characteristics of a dataset through statistical methods and visualization.

We will be using the big-mac pricing dataset we imported earlier as an example for the following sections.

3.1 Data cleaning and selection

Data cleaning and selection are typically the initial steps in EDA. In practice, most datasets you encounter will be large and may contain issues such as typos, missing values, or other errors.

3.1.1 Cleaning the dataframe

We can first start by cleaning all the messy data. This includes Null/NA values, misinputs, and in some cases, extreme values. There are multiple approaches for this, such as deleting the data or imputation (replacing them with median or mean), etc.

In this case, we will use the dplyr package to drop all the Null/NA values.

"Data_BigMac <- Data_BigMac %>% drop_na()"

We can also use mean or median to replace any missing or extreme values. Here is an example code snippet for you to get familiar with the syntax.

Data_BigMac$dollar_ex[is.na(Data_BigMac$dollar_ex)] <- mean(Data_BigMac$dollar_ex, na.rm = T) 
Data_BigMac$dollar_ex[is.na(Data_BigMac$dollar_ex)] <- median(Data_BigMac$dollar_ex, na.rm = T)

We can also use the mice package to help us with more advanced imputations. Here is the syntax for a mice imputation function.

"head(nhanes)"
"imputed_data <- mice(nhanes, m = 5, method = 'pmm', print = FALSE)"
"head(complete(imputed_data, action = 1))"

3.1.2 Select the sections of dataframe

When we are working with a large dataframe, it is standard procedure to drop unnecessary data to save memory space.

We will be using the filter() function from the dplyr package in this example. Let’s filter out any country that has a dollar exchange rate lower than 2.

Data_BigMac <- filter(Data_BigMac, dollar_ex >= 2)
head(Data_BigMac)
        date currency_code           name local_price dollar_ex dollar_price
1 2000-04-01           CLP          Chile     1260.00    514.00     2.451362
2 2000-04-01           CNY          China        9.90      8.28     1.195652
3 2000-04-01           CZK Czech Republic       54.37     39.10     1.390537
4 2000-04-01           DKK        Denmark       24.75      8.04     3.078358
5 2000-04-01           HKD      Hong Kong       10.20      7.79     1.309371
6 2000-04-01           HUF        Hungary      339.00    279.00     1.215054

3.2 Data manipulation and statistical summary

3.2.1 Common functions to inspect a dataframe

It is routine practice to take a look at the actual data in the dataframe to have a better understanding of what we are dealing with. Functions like head() and tail() are a good start. You can also use dim() to know how many rows and columns there are for this dataframe.

head(Data_BigMac)
        date currency_code           name local_price dollar_ex dollar_price
1 2000-04-01           CLP          Chile     1260.00    514.00     2.451362
2 2000-04-01           CNY          China        9.90      8.28     1.195652
3 2000-04-01           CZK Czech Republic       54.37     39.10     1.390537
4 2000-04-01           DKK        Denmark       24.75      8.04     3.078358
5 2000-04-01           HKD      Hong Kong       10.20      7.79     1.309371
6 2000-04-01           HUF        Hungary      339.00    279.00     1.215054
tail(Data_BigMac)
           date currency_code         name local_price   dollar_ex dollar_price
1027 2020-01-14           TRY       Turkey       12.99     5.88025     2.209090
1028 2020-01-14           TWD       Taiwan       72.00    29.88450     2.409276
1029 2020-01-14           UAH      Ukraine       57.00    23.99000     2.375990
1030 2020-01-14           UYU      Uruguay      179.00    37.43500     4.781621
1031 2020-01-14           VND      Vietnam    66000.00 23176.00000     2.847774
1032 2020-01-14           ZAR South Africa       31.00    14.39100     2.154124
dim(Data_BigMac)
[1] 1032    6
summary(Data_BigMac)
     date           currency_code          name            local_price     
 Length:1032        Length:1032        Length:1032        Min.   :      2  
 Class :character   Class :character   Class :character   1st Qu.:     15  
 Mode  :character   Mode  :character   Mode  :character   Median :     49  
                                                          Mean   :  13356  
                                                          3rd Qu.:    280  
                                                          Max.   :4000000  
   dollar_ex          dollar_price  
 Min.   :      2.0   Min.   :0.640  
 1st Qu.:      5.8   1st Qu.:2.239  
 Median :     18.6   Median :2.743  
 Mean   :   5081.9   Mean   :3.011  
 3rd Qu.:    102.0   3rd Qu.:3.591  
 Max.   :1600500.0   Max.   :8.312

3.2.2 Statistical summary of the dataframe

Statistical summary is another beneficial way to understand the data. We can use some built-in functions in R.

mean(Data_BigMac$local_price)
[1] 13356.14
median(Data_BigMac$dollar_price)
[1] 2.742555
max(Data_BigMac$dollar_ex)
[1] 1600500
min(Data_BigMac$dollar_ex)
[1] 2.01
sd(Data_BigMac$local_price)
[1] 210191.9
var(Data_BigMac$dollar_ex)
[1] 6443729619

3.2.3 A simple regression example

Regression is one of the simplest and widely used data analysis tools. We will run an easy regression using the base R lm() function in the following example.

Note: This example is only to get you familiar with the syntax for running a regression, the result will not make sense at all!

model <- lm(Data_BigMac$local_price ~ Data_BigMac$dollar_price, Data_BigMac)
coef(model)   
             (Intercept) Data_BigMac$dollar_price 
               22881.616                -3163.425 
summary(model)

Call:
lm(formula = Data_BigMac$local_price ~ Data_BigMac$dollar_price, 
    data = Data_BigMac)

Residuals:
    Min      1Q  Median      3Q     Max 
 -20855  -15566  -13547   -9406 3986672 

Coefficients:
                         Estimate Std. Error t value Pr(>|t|)
(Intercept)                 22882      17811   1.285    0.199
Data_BigMac$dollar_price    -3163       5501  -0.575    0.565

Residual standard error: 210300 on 1030 degrees of freedom
Multiple R-squared:  0.0003209, Adjusted R-squared:  -0.0006496 
F-statistic: 0.3307 on 1 and 1030 DF,  p-value: 0.5654

3.3 Data Visualization

Data visualization is another essential aspect of EDA. It is especially useful when you present your work to a larger audience.

In the following sections, we will go through a few prevalent graphs. We will be using the ggplot2 package in the example code.

3.3.1 Scatter plot

Scatter plot is best applied when you want to display the relationship between two sets of data. Here is an example dataframe and graph using ggplot2.

Gwu_Student_HW <- data.frame(
  height = c(177, 180, 190, 176, 161, 143, 189, 167, 183, 185),   
  weight = c(157, 160, 207, 147, 180, 115, 220, 150, 183, 190),    
  gender = c('male','male','female','female','male','male','female','female','female','male')
)
head(Gwu_Student_HW)
  height weight gender
1    177    157   male
2    180    160   male
3    190    207 female
4    176    147 female
5    161    180   male
6    143    115   male
plot(
  Gwu_Student_HW$height,
  Gwu_Student_HW$weight,
  pch = ifelse(Gwu_Student_HW$gender == 'male', 16, 17),
  col = ifelse(Gwu_Student_HW$gender == 'male', 'blue', 'red'),
  xlab = 'Height (cm)',
  ylab = 'Weight (lbs)',
  main = 'Height vs Weight (blue = male, red = female)'
)

3.3.2 Bar Chart

Bar chart is typically used for categorical data. Here is an example using base R.

Average_Rain_Volumn_by_Seasons <- data.frame(
  season = c("Spring", "Summer", "Fall", "Winter"),
  volumns = c(76, 33, 56, 21)
)
barplot(Average_Rain_Volumn_by_Seasons$volumns, 
names.arg = Average_Rain_Volumn_by_Seasons$season,
        col = "steelblue", main = "Average Rain Volumn by Seasons",
        xlab = "seasons", ylab = "average rain volumns")

3.3.3 Boxplot

Boxplot is a great way to show how data is distributed. Here is an example using ggplot2.

GWU_Econ_Semester_StudenGPA <- data.frame(
  group = rep(c("Spring2024", "Fall2024", "Spring2025"), each = 20),
  value = c(
    rnorm(20, mean = 3.6, sd = 0.2),
    rnorm(20, mean = 3.4, sd = 0.3),
    rnorm(20, mean = 3.5, sd = 0.1)
  )
)
head(GWU_Econ_Semester_StudenGPA)
       group    value
1 Spring2024 3.853097
2 Spring2024 3.620653
3 Spring2024 3.585771
4 Spring2024 3.835464
5 Spring2024 3.499387
6 Spring2024 3.403021
boxplot(
  value ~ group,
  data = GWU_Econ_Semester_StudenGPA,
  col = c("skyblue", "palegreen3", "lightpink"),
  notch = TRUE,
  main = "GWU Econ Student GPA by Semester",
  xlab = "Semester",
  ylab = "GPA",
  cex.lab = 1.1,
  cex.axis = 0.9
)
Warning in (function (z, notch = FALSE, width = NULL, varwidth = FALSE, : some
notches went outside hinges ('box'): maybe set notch=FALSE

4 Interview Questions / Quiz Yourself!

NoteNote

On this page, you will find some interview questions to practice your advanced knowledge of R programming. All of the content is based on the R Advanced tutorial from this page.

Even if you are not preparing for an interview, these questions can help test your knowledge (perhaps for an upcoming exam, certification, or homework assignment!).

TipTip

Try to actively recall the answers before checking them. This is one of the most effective ways to learn and increase retention!

Download Advanced R Quiz

NoteQuestion 1: Which is NOT a function apart of base R?

A. dim(students)
B. var(movie_lengths)
C. drop_na(height)
D. is.na(homescore)

C. drop_na() - This comes from the dplyr package!

NoteQuestion 2: Which of the following correctly selects a section of the dataframe tickets_sold?

tickets_sold <- ???

A. filter(tickets_sold, dollar_spent >= 150)
B. select(tickets_sold, dollar_spent$150)
C. filter(tickets_sold, dollar_spent === 150)
D. select(tickets_sold, dollar_spent{150})

A. filter(tickets_sold, dollar_spent >= 150)

NoteQuestion 3: Which argument makes a scatterplot within ggplot?

MAYBE: Which of the following needs the + operator between components of the argument? A. geom_plot()
B. geom_point()
C. geom_splot()
D. geom_scatter()

B. geom_point()

NoteQuestion 4: Which of the following correctly creates a boxplot using the data frame intern_gpa?

Assume intern_gpa contains two columns:

cohort (“Spring”, “Fall”, “Summer”) gpa (numeric values)

A. ggplot(intern_gpa, aes(x = cohort, y = gpa)) + geom_boxplot(), labs(title = “Boxplot of Intern Cohort GPAs”, x = “Cohort Season”, y = “GPA”)
B. ggplot(intern_gpa + aes(x = cohort, y = gpa)) + geom_boxplot() + labs(title = “Boxplot of Intern Cohort GPAs”, x = “Cohort Season”, y = “GPA”)
C. ggplot(intern_gpa, aes(x = cohort, y = gpa)), geom_boxplot() + labs(title = “Boxplot of Intern Cohort GPAs”, x = “Cohort Season”, y = “GPA”)
D. ggplot(intern_gpa, aes(x = cohort, y = gpa)) + geom_boxplot() + labs(title = “Boxplot of Intern Cohort GPAs”, x = “Cohort Season”, y = “GPA”)

D. ggplot(intern_gpa, aes(x = cohort, y = gpa)) + geom_boxplot() + labs(title = “Boxplot of Intern Cohort GPAs”, x = “Cohort Season”, y = “GPA”)

NoteQuestion 5: What needs to be fixed in this code snippet to avoid errors? 
cafe_q1sales <- data.frame(
  total_sales = rnorm(100, mean = 1200, sd = 300),   
  transactions = sample(50:100, 50, replace = TRUE), 
  gender = sample(c("Male", "Female"), 50, replace = TRUE)
)
ggplot(cafe_q1sales, aes(x = transactions, y = total_sales, color = gender)) +
  geom_point(size = 5) +
  labs(
    title = "Cafe Q1 Sales by Transaction and Gender",
    x = "Number of Transactions",
    y = "Total Sales ($)"
  )

A. Need to specify cafe_q1sales to use when making the scatterplot.
B. Labels don’t need quotation marks around their titles.
C. Need another comma here: gender = sample(c(“Male”, “Female”), 50, replace = TRUE),
D. Replace must be false, not true.

A. Need to specify cafe_q1sales to use when making the scatterplot.

4.1 Bonus: Create your own bar chart based off of this data frame!

school_acceptance <- data.frame(
  school = c("Yale", "GWU", "Georgetown", "UVA"),
  acceptance_ratepercent = c(4.5, 43.5, 13.1, 16.9)
)
Back to top