For example, you may need to use the function lm.gls() implemented in the package MASS to fit a linear model using the generalized least squares method.
library(MASS) #load the package MASS help(lm.gls) #get help on this function. Same as typing ?lm.gls
Getting started
One problem that we usually face when we load and attach some libraries in R is that these libraries might have different masked functions share the same namespace. For example, the function lag() is masked by both stats and dplyr packages. It performs a different tasks in both. Thus, you need to be careful if you are using lag() in R while the package dplyr.
stats::lag #to explicitly use the lag() function from the stats packagedplyr::lag #to explicitly use the lag() function from the dplyr package
Example
set.seed(1) #set reproducible results x <-rnorm(5) #generate 5 observations from the standard normal distribution N(0,1)stats::lag(x, 2) #shift the time base back by 2 (keep 1st & 2nd observations)
dplyr::lag(x, 2) #shift the time base back by 2 (replace 1st & 2nd observations)
[1] NA NA -0.6264538 0.1836433 -0.8356286
Simple R code
Let’s start with the following R code
# create two numeric vectors, each with 8 observationswt <-c(60,70,63,55,48,49,58,58)age =c(20,17,23,24,19,19,16,26) #note "<-" symbol can be replaced by "="# get a random sample (without replacement) of 8 observationsset.seed(5) #set seed to reproduce the same random samplez=sample(150:190, size =8)# replicat a string "Male" 3 times & get a vector of characters Male=rep("Male", times =3)# replicat a string "Female" 5 times & get a vector of characters Female=rep("Female", times =5)# combine the two categorical variables into one nominal variables =c(Male, Female)# create an ordinal categorical variable income=c("Low","High","Low","Low","Middle","Middle","Middle","High")# stores categorical values as vector of integers (factors)sex=factor(s)income=factor(income,order=TRUE,levels=c("Low","Middle","High"))# create a data frame and name the variablesmydata=data.frame(id=1:8,weight=wt,age=age,z=z,sex=sex,Sex=s,income)
# Create a new datasetstatus =c("worker","student","worker","worker","student","student","student","student")my_newdata <-data.frame(id=c(1:4,8:11),status=status)
#return first 6 observationshead(mydata)
id weight age z sex Sex income
1 1 60 20 151 Male Male Low
2 2 70 17 164 Male Male High
3 3 63 23 160 Male Male Low
4 4 55 24 170 Female Female Low
5 5 48 19 179 Female Female Middle
6 6 49 19 156 Female Female Middle
#get variables namesnames(mydata)
[1] "id" "weight" "age" "z" "sex" "Sex" "income"
#display the Structure of the datastr(mydata)
'data.frame': 8 obs. of 7 variables:
$ id : int 1 2 3 4 5 6 7 8
$ weight: num 60 70 63 55 48 49 58 58
$ age : num 20 17 23 24 19 19 16 26
$ z : int 151 164 160 170 179 156 168 152
$ sex : Factor w/ 2 levels "Female","Male": 2 2 2 1 1 1 1 1
$ Sex : chr "Male" "Male" "Male" "Female" ...
$ income: Ord.factor w/ 3 levels "Low"<"Middle"<..: 1 3 1 1 2 2 2 3
mat1 <-matrix(wt, nrow =5, ncol =2) #create a matrix of dimension 5x2mat2 <-matrix(age, nrow =2, ncol =5) #create a matrix of dimension 2x5mat3 =cbind(wt,age,sex,income) #combine vectors by columns. Exercise: Type the code: rbind(wt,age,sex,income) and explain the outcome!mylist <-list(wt,age,sex,income) #create a list of 4 vectors
dim(mydata) # dimension
[1] 8 7
dim(mat1)
[1] 5 2
dim(mat3)
[1] 8 4
length(mylist)
[1] 4
class(wt)
[1] "numeric"
class(sex)
[1] "factor"
class(s)
[1] "character"
class(mydata)
[1] "data.frame"
class(mat1)
[1] "matrix" "array"
class(mylist)
[1] "list"
income[1:4] #first 4 elements of income
[1] Low High Low Low
Levels: Low < Middle < High
wt[-2] #all weights without 2nd one
[1] 60 63 55 48 49 58 58
mylist[1] #elements of the 1st list
[[1]]
[1] 60 70 63 55 48 49 58 58
mylist[[1]][2] #2nd element of the 1st list
[1] 70
mydata$income #values of the income variable
[1] Low High Low Low Middle Middle Middle High
Levels: Low < Middle < High
mydata[2,] #values of the 2nd row
id weight age z sex Sex income
2 2 70 17 164 Male Male High
mydata[, 4] #values of the 4th column
[1] 151 164 160 170 179 156 168 152
# Extract a subset data# rows 1,2,3,5,7 and columns id, weight, sexmydata[c(1:3,5,7),c("id","weight","sex")]
id weight sex
1 1 60 Male
2 2 70 Male
3 3 63 Male
5 5 48 Female
7 7 58 Female
#value in 3rd row and 2nd columnmydata[3, 2]
[1] 63
# Merging datasetsmerge(my_newdata,mydata, by ="id")
id status weight age z sex Sex income
1 1 worker 60 20 151 Male Male Low
2 2 student 70 17 164 Male Male High
3 3 worker 63 23 160 Male Male Low
4 4 worker 55 24 170 Female Female Low
5 8 student 58 26 152 Female Female High
mean(mydata$weight) #mean weight
[1] 57.625
#standard deviation of weight variablesd(mydata[,"weight"])
[1] 7.190023
#Pearson correlation coefficient between weight and agecor(mydata[,2],mydata[,"age"])
id weight age z sex
Min. :1.00 Min. :48.00 Min. :16.00 Min. :151.0 Female:5
1st Qu.:2.75 1st Qu.:53.50 1st Qu.:18.50 1st Qu.:155.0 Male :3
Median :4.50 Median :58.00 Median :19.50 Median :162.0
Mean :4.50 Mean :57.62 Mean :20.50 Mean :162.5
3rd Qu.:6.25 3rd Qu.:60.75 3rd Qu.:23.25 3rd Qu.:168.5
Max. :8.00 Max. :70.00 Max. :26.00 Max. :179.0
Sex income
Length:8 Low :3
Class :character Middle:3
Mode :character High :2
# cross-product of two matrices# try: crossprod(mat2,t(mat1))mat1 %*% mat2
# Creating new variablesmydata$status0 <-paste0("grade",1:8)mydata
id weight age z sex Sex income status0
1 1 60 20 151 Male Male Low grade1
2 2 70 17 164 Male Male High grade2
3 3 63 23 160 Male Male Low grade3
4 4 55 24 170 Female Female Low grade4
5 5 48 19 179 Female Female Middle grade5
6 6 49 19 156 Female Female Middle grade6
7 7 58 16 168 Female Female Middle grade7
8 8 58 26 152 Female Female High grade8
# Creating new variablesmydata$status1 <-paste("grade",1:8)mydata
id weight age z sex Sex income status0 status1
1 1 60 20 151 Male Male Low grade1 grade 1
2 2 70 17 164 Male Male High grade2 grade 2
3 3 63 23 160 Male Male Low grade3 grade 3
4 4 55 24 170 Female Female Low grade4 grade 4
5 5 48 19 179 Female Female Middle grade5 grade 5
6 6 49 19 156 Female Female Middle grade6 grade 6
7 7 58 16 168 Female Female Middle grade7 grade 7
8 8 58 26 152 Female Female High grade8 grade 8
# Recoding variables: recode age 20 by a missing valuemydata$age[mydata$age ==20] <-NAmydata[1:4,] #display the first 4 rows
id weight age z sex Sex income status0 status1
1 1 60 NA 151 Male Male Low grade1 grade 1
2 2 70 17 164 Male Male High grade2 grade 2
3 3 63 23 160 Male Male Low grade3 grade 3
4 4 55 24 170 Female Female Low grade4 grade 4
#calculate the mean including missing valuessum(mydata$age)
[1] NA
#calculate the mean excluding missing valuessum(mydata$age, na.rm=TRUE)
[1] 144
# Renaming variablesnames(mydata)[4] <-"height"
# Dropping variablesmydata[,-c(4,9)] #Dropping the variables "height" and "status1"
id weight age sex Sex income status0
1 1 60 NA Male Male Low grade1
2 2 70 17 Male Male High grade2
3 3 63 23 Male Male Low grade3
4 4 55 24 Female Female Low grade4
5 5 48 19 Female Female Middle grade5
6 6 49 19 Female Female Middle grade6
7 7 58 16 Female Female Middle grade7
8 8 58 26 Female Female High grade8
# Removing all rows with missing datanewdata <-na.omit(mydata)
par(mfrow =c(2, 2)) #create a 2 x 2 plotting matrixplot(wt,age); plot(mydata$weight, mydata$age) #type ?plot to get help about the function plot()plot(wt,age, xlab ="Weight", ylab ="Age", col ="red")plot(density(rnorm(500)),col="blue") #plot a density distribution of 500 random data from Gaussian
Reading data into R
After setting up R environment with Rstudio, you can import the data from different structures.
The build in R package utils has several functions for reading files from delimited ASCII (.txt, .csv) files.
read.table(): Import data with extension .txt.
read.csv(): Import data where “,” are used as separators and “.” are used as decimals.
read.csv2(): Import data where “;” are used as separators and “,” are used as decimals.
> read.csv
function (file, header = TRUE, sep = ",", quote = "\"", dec = ".", fill = TRUE, comment.char = "", ...)
> read.csv2
function (file, header = TRUE, sep = ";", quote = "\"", dec = ",", fill = TRUE, comment.char = "", ...)
Reading data into R: Example
Bitcoin daily price (in US dollars) from January 22, 2020 to September 1, 2021 (during COVID-19).
# change the working directory to a different location on your computerdat1 <-read.table("data/BTC.txt",header = T, fill=TRUE) dat2 <-read.csv("data/BTC.csv", header =TRUE)str(dat1)
'data.frame': 589 obs. of 5 variables:
$ Date : chr "1/22/2020" "1/23/2020" "1/24/2020" "1/25/2020" ...
$ Price: num 8664 8404 8447 8354 8622 ...
$ Open : num 8734 8669 8404 8447 8351 ...
$ High : num 8800 8669 8522 8447 8622 ...
$ Low : num 8581 8297 8248 8280 8304 ...
Use skim() function from skimr package to get a useful summary statistics.
#first, make sure you that have the package "readr" installed on you computer,#if not installed, you need to install itinstall.packages("skimr") library("skimr") skim(dat1)
Exercise:
Explain the arguments row.names, col.names, stringsAsFactors in the function read.table().
Give examples to demonstrate how we can use these functions.
Reading data into R
The readr package provides functions to read rectangular data with extension .csv, .txt or .tsv.
read_csv() and read_tsv() are special cases of the more general read_delim().
To see all the arguments for the function read_csv(), use ?read_csv() within R.
By default, read_csv() and read_tsv() convert blank cells to missing data (NA).
You can type the following command to install and load this package.
#first, make sure you that have the package "readr" installed on you computer,#if not installed, you need to install itinstall.packages("readr")library(readr) # Load the package
# Read from a path specifies the location of a data on your computername_data <-read_csv("file_data - Sheet1.csv") # import data from a comma delimited file# Read from a remote path (e.g., mtcars data set from GitHub website)name_data <-read_csv("https://github.com/tidyverse/readr/raw/main/inst/extdata/mtcars.csv") name_data <-read_tsv("file_data.txt") # import data from a tab delimited file separated by tabsname_data <-read_tsv("file_data.tsv", sheet=1) # import data from a tab delimited file
Reading data into R
The readxl package can import tabular data from Excel workbooks. Both xls and xlsx formats are supported.
#first, make sure you that have the package "readxl" installed on you computer,#if not installed, you need to install itinstall.packages("readxl") library(readxl) # Load the packagename_data <-read_excel("file_data.xlsx", sheet=1) # import data from an Excel workbook
The haven package can import data with .sav, .dat, and .sas7bdat extensions.
#first, make sure you that have the package "haven" installed on you computer,#if not installed, you need to install itinstall.packages("haven") library(haven) # Load the packagename_data <-read_sav("file_data.sav") # import data from SPSSname_data <-read_dat("file_data.dat") # import data from Stataname_data <-read_sas("file_data.sas7bdat") # import data from SAS
Example (reading data into R):
Import the BTC data set using the functions read_tsv() and read_csv() from the package readr.
library(readr)dat3 <-read_tsv("data/BTC.txt") dat4 <-read_csv("data/BTC.csv")head(dat3) #same results using head(dat4)
Note that the head() prints differently from before because it’s a tibble. Tibbles are rectangular data frames, but slightly tweaked to work better in the tidyverse package that we will discuss later!
The excel sheet BTC2 has two sheets named BTC and BTC2. The data BTC2 is stored in columns G7:G38-K7:38. The first cell A1 provides a quick description of this data. Data has some missing values.
# A tibble: 6 × 5
Date Price Open High Low
<dttm> <dbl> <dbl> <dbl> <dbl>
1 2021-01-01 00:00:00 29346 28933 29498 28932
2 2021-01-02 00:00:00 32185 29346 33168 29192
3 2021-01-03 00:00:00 32971 32183 34253 32110
4 2021-01-04 00:00:00 NA NA NA NA
5 2021-01-05 00:00:00 33996 32020 33996 30979.
6 2021-01-06 00:00:00 36755 33986 36755 33901
Tibbles for tidy data frames
Many popular packages, such as readr, tidyr, dplyr, and purr, save data frames as tibbles. When you are using the package tibble to import data be aware of the following properties:
Data frames can be converted to tibbles using the function as_tibble().
library(tibble)class(mtcars) #the class of mtcars data before tibble
[1] "data.frame"
mtcars <-as_tibble(mtcars) class(mtcars) #the class of mtcars data after tibble
[1] "tbl_df" "tbl" "data.frame"
tibbles never change the names of variables.
tibbles never convert character variables to factors.
tibbles don’t support row names.
Subsetting a tibble always returns a tibble and not a vector.
In financial time series, returns are given by r_t =\log\dfrac{P_t}{P_{t-1}}, where r_t and P_t denote the returns and price asset at time t.
library(dplyr)dat5 <-mutate(dat5,BeforeClose=dplyr::lag(Price), returns=log(Price)-log(BeforeClose)) #use the BTC2 datasetdat5[1:2,]
# A tibble: 2 × 7
Date Price Open High Low BeforeClose returns
<dttm> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 2021-01-01 00:00:00 29346 28933 29498 28932 NA NA
2 2021-01-02 00:00:00 32185 29346 33168 29192 29346 0.0923
Use the pipe operator (%>%) to chain statements.
dat5 <- dat5 %>%mutate(Date = lubridate::mdy(Date), #parse dates with month, day, and year components using the function mdy() from the "lubridate" packageBeforeClose=dplyr::lag(Price),returns=log(Price)-log(BeforeClose))dat5[1:2,] #note that the first returns is missing "NA". To remove "NA", use the code: %>% tidyr::drop_na()
# A tibble: 2 × 7
Date Price Open High Low BeforeClose returns
<date> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 NA 29346 28933 29498 28932 NA NA
2 NA 32185 29346 33168 29192 29346 0.0923
# Select the variables id, weight, sex, and incomemydata %>%select(id,weight,sex,income)
id weight sex income
1 1 60 Male Low
2 2 70 Male High
3 3 63 Male Low
4 4 55 Female Low
5 5 48 Female Middle
6 6 49 Female Middle
7 7 58 Female Middle
8 8 58 Female High
# A minus sign (-) is used to exclude variablesmydata %>%select(-height, -Sex,-status1)
id weight age sex income status0
1 1 60 NA Male Low grade1
2 2 70 17 Male High grade2
3 3 63 23 Male Low grade3
4 4 55 24 Female Low grade4
5 5 48 19 Female Middle grade5
6 6 49 19 Female Middle grade6
7 7 58 16 Female Middle grade7
8 8 58 26 Female High grade8
# Select all females with age 19 or weight greater than 59mydata %>%filter(sex =="Female"& age ==19| weight >59)
id weight age height sex Sex income status0 status1
1 1 60 NA 151 Male Male Low grade1 grade 1
2 2 70 17 164 Male Male High grade2 grade 2
3 3 63 23 160 Male Male Low grade3 grade 3
4 5 48 19 179 Female Female Middle grade5 grade 5
5 6 49 19 156 Female Female Middle grade6 grade 6
# A tibble: 3 × 2
income avg_age
<ord> <dbl>
1 Low NA
2 Middle 18
3 High 21.5
Note that the first age is missing (“NA”). This value is associated with low income. Thus, the average age for those who have low income is missing (“NA”).
id age height weight gender income status height_foot
1 6 19 156 49 Female Middle grade6 5.148
2 3 23 160 63 Male Low grade3 5.280
Exercise:
Use the functions group_by() to group data by status and calculate the five numbers (min, max, median, first and third quartiles) of each status.
Group data by status and income and calculate the mean and standard deviation of each group based on their age.
Probability functions in R
In many scenarios we use probability functions to generate a simulated data.
In R, probability functions take the form [dpqr] abbreviation name of distribution, where each letter of [dpqr] refers to the aspect of the distribution returned:
d = Density
p = Distribution function
q = Quantile function
r = Random generation
Distribution
Syntax
Distribution
Syntax
Distribution
Syntax
Beta
beta()
Binomial
binom()
Cauchy
cauchy()
Chi-squared
chisq()
Exponential
exp()
F
f()
Gamma
gamma()
Geometric
geom()
Hypergeometric
hyper()
Lognormal
lnorm()
Logistic
logis()
Multinomial
multinom()
Negative binomial
nbinom()
Normal
norm()
Poisson
pois()
Wilcoxon signed rank
signrank()
T
t()
Uniform
unif()
Weibull
weibull()
Wilcoxon rank sum
wilcox()
Example:
Q1: What is the area under the standard normal curve to the left of z = 2.1?
Answer: Use the command pnorm(2.1) to get 0.9821356.
Q2: What is the value of the 95th percentile of a normal distribution with a mean of 100 and a standard deviation of 20?
Answer: Use the command qnorm(0.95, mean =100, sd = 20) to get 132.8971
Q3: Generate 300 random normal deviates with a mean of 80 and a standard deviation of 10.
Answer: Use the command rnorm(300, mean =80, sd = 10) to get the simulated series.
You may install and load the tidyverse by running the following code:
> install.packages("tidyverse")
> library(tidyverse)
── Attaching core tidyverse packages ─────────────────────────────────────────── tidyverse 2.0.0 ──
✔ dplyr 1.1.2 ✔ readr 2.1.4
✔ forcats 1.0.0 ✔ stringr 1.5.0
✔ ggplot2 3.4.2 ✔ tibble 3.2.1
✔ lubridate 1.9.2 ✔ tidyr 1.3.0
✔ purrr 1.0.1
── Conflicts ───────────────────────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag() masks stats::lag()
ℹ Use the conflicted package to force all conflicts to become errors
Warning messages:
1: package ‘tidyverse’ was built under R version 4.3.1
2: package ‘readr’ was built under R version 4.3.1
A graphing template
The ggplot2 implements the grammar of graphics is a core package in the tidyverse that can be used to visualize data in an elegant way.
To get started plotting in ggplot2, you will start with the two basics (data and a geom) and then add additional layers using the + sign.
library(tidyverse) # or type library(ggplot2) to load the package ggplot2 onlyggplot(data = DATAFRAME, mapping =aes(Options)) +geom_TYPE()## You may place the mapping within the aes function, where aes stands for aesthetics.ggplot(data = DATAFRAME) +geom_TYPE(mapping =aes(Options))
Each geom_ function takes a mapping argument which paired with aes().
The x and y arguments of aes() specify which variables options to map to the x and y coordinates.
You can add a third variable, like z and map it to an aesthetic like the size, shape, or color/colour of the points of your plot.
Let’s use our first graph to answer the following questions about the mpg data frame available from the package ggplot2:
Do cars with big engines use more fuel than cars with small engines?
What does the relationship between engine size and fuel efficiency look like?
Is it positive or negative?
Is it Linear or nonlinear?
head(mpg) # Returns the first 6 rows
# A tibble: 6 × 11
manufacturer model displ year cyl trans drv cty hwy fl class
<chr> <chr> <dbl> <int> <int> <chr> <chr> <int> <int> <chr> <chr>
1 audi a4 1.8 1999 4 auto(l5) f 18 29 p compa…
2 audi a4 1.8 1999 4 manual(m5) f 21 29 p compa…
3 audi a4 2 2008 4 manual(m6) f 20 31 p compa…
4 audi a4 2 2008 4 auto(av) f 21 30 p compa…
5 audi a4 2.8 1999 6 auto(l5) f 16 26 p compa…
6 audi a4 2.8 1999 6 manual(m5) f 18 26 p compa…
Creating a ggplot - static plots
# Creates a coordinate system that you can add layers to ggplot(data = mpg)
The first argument is the dataset that you need to use in the plot. The result of this code is an empty graph (default theme used by ggplot2 is theme_gray()).
Now you can add one or more layers to ggplot(). The function geom_point() adds a layer of points (scatterplot) to your plot.
# Put displ on the x-axis and hwy on the y-axis ggplot(data = mpg) +geom_point(mapping =aes(x = displ, y = hwy))
The plot shows a negative relationship between the car engine size (in liters) and the car’s fuel efficiency on the highway (in miles per gallon). The bigger the size of the engine, the less efficient it is in consuming fuel.
Aesthetic mappings
The aes() (stands for aesthetics) function is used to map variables to the visual characteristics of a plot.
# Map the colors of points to the variable "class", which indicates the class of each carggplot(data = mpg) +geom_point(mapping =aes(x = displ, y = hwy, color = class))
The graph indicates that there are some unusual points (outliers). The colors reveal that many of these points are two-seater cars.
Map an aesthetic to a third variable will create a legend that explains which levels correspond to which values.
Aesthetic mappings
# Map the size of points to the variable "class"ggplot(data = mpg) +geom_point(mapping =aes(x = displ, y = hwy, size = class))
Mapping class to the size aesthetic display more clear information about the (outliers) in this data.
Aesthetic mappings
# Map the variable "class" to the alpha aesthetic# alpha controls the transparency of the points or the shape of the points. ggplot(data = mpg) +geom_point(mapping =aes(x = displ, y = hwy, alpha = class))
The visualization of the plot is not clear. We can fine tune the appearance of the graph using themes and improved visualization.
Aesthetic mappings
# Map the variable "class" to the shape, color, and size aesthetics and use the classic themeggplot(data = mpg) +geom_point(mapping =aes(x = displ, y = hwy, shape = class, colour = class, size = class)) +theme_classic()
ggplot2 uses only six shapes at a time. Thus, the suv go unplotted when you use the aesthetic shape.
Both British and American English words of colour and color work.
Exercise: Is it possible to replace the argument (e.g., color) in R by an abbreviations word using the first character strings of this argument (e.g., col)?
Set the aesthetic properties of the geom manually
Here, I will use the theme theme_bw() (for black and white).
ggplot(data = mpg) +geom_point(aes(displ,hwy,col=drv),size =4) #a legend to the right will be created
Scales
Scale functions (which start with scale_) allow you to modify default scaling provided by ggplot2
Function
Syntax
scale_x_continuous()
Scales the x-axis for quantitative variables. Options include breaks for specifying tick marks, labels for specifying tick mark labels, and limits to control the range of the values displayed
scale_y_continuous()
Same as above for y-axis
scale_x_discrete()
Same as above for x-axis representing categorical variable
scale_y_discrete()
Same as above for y-axis representing categorical variable
scale_color_manual()
Specifies the colors (with option values) used to represent the levels of a categorical variable
Facets
facet_wrap() and facet_grid() are used to partition a plot into a matrix of panels (side-by-side graphs), particularly useful for categorical variables.
Function
Syntax
facet_wrap(~var, nrow = r)
Partition plots for each level of variable (var) arranged into r rows
facet_wrap(~var, ncol = c)
Partition plots for each level of variable (var) arranged into c columns
facet_grid(row_var~col_var)
Partition plots for combination of rows variable (row_var) and columns variable (col_var)
facet_grid(rows = row_var)
Partition plots for for each level of rows variable (row_var), arranged as a single column
facet_grid(cols = col_var)
Partition plots for for each level of columns variable (col_var), arranged as a single row
# Map the variable "class" to the shape, color, and size aesthetics and use the classic themeggplot(data = mpg) +geom_point(mapping =aes(x = displ, y = hwy)) +facet_wrap(~ class, nrow =2) +theme_bw()
To facet your plot by a single variable, use facet_wrap( ~ name of this variable).
Note that the variable that you pass to facet_wrap() should be discrete.
Exercise: What happens if you replace facet_wrap(~ class, nrow = 2) by facet_grid(~ class)?
# Bar graphs ggplot(data=mpg,mapping=aes(x=hwy))+geom_bar() +facet_grid(rows =vars(drv), margins =TRUE, scales ="free_y")
Note: The default argument scales = “fixed” is used if x and y scales are fixed across all panels; scales = “free_x” if x scale is free and y scale is fixed; scales = “free_y” if y scale is free and x scale is fixed; and scales = “free” if x and y scales vary across panels.
Facets: Facet the plot on the combination of two variable
# Map the variable "class" to the shape, color, and size aesthetics and use the classic themeggplot(data = mpg) +geom_point(mapping =aes(x = displ, y = hwy)) +facet_grid(drv ~ cyl) +theme_bw()
To facet your plot on the combination of two variable, use facet_grid(name of 1st variable ~ name of 2nd variable).
Here, 4 stands for four-wheel drive, f for front-wheel drive, and r for rear-wheel drive.
Legend layout
To control the position of the legend, you need to use a theme() setting.
The theme setting legend.position controls where the legend is drawn:
+ theme(legend.position = "right") # the default
+ theme(legend.position = "left")
+ theme(legend.position = "top")
+ theme(legend.position = "bottom")
You may include other options in the theme() as seen below!
suv <- mpg %>%filter(class =="suv")p <-ggplot(suv, aes(displ, hwy, color = drv)) +geom_point(size =4) +theme_bw()p +labs(title ="Fuel economy data",subtitle ="Suv cars",x ="Engine displacement, in litres",y ="Highway miles per gallon",color ="Type of drive train") +scale_color_manual(labels =c("4wd", "Rear wheel drive"), values =c("blue", "red")) +theme(legend.position="bottom", legend.key.size =unit(1.4, "cm"),legend.key.height=unit(0.5, "cm"),legend.key =element_rect(fill ="gray90", color ="red"),text=element_text(family="serif"))
You can map the values of a non categorical variable to a different continuous scale using the scale_color_gradient() option
ggplot(data = mpg) +geom_point(aes(displ, hwy,col=year),size =4) +scale_color_gradient(low ="green", high ="purple")
You can change the overall theme.
A list of themes and instructions can be found from:
Helps visualize whether a distribution of a data set is symmetric or skewed due to unusual observations (outliers). The grapgh displays the five numbers summary (minimum, maximum, median, first and third quartiles).
Use the function tq_get() available from the package tidyquant to get stock price data (say Apple} from several web sources.
Define the returns as a time series data and plot this series using line chart.
Plot the distribution using histogram and annotate the plots.
Use the function grid.arrange() available from the package gridExtra to place the two plots on one page.
Use the option annotate() in ggplot2 to add a text label to your plot.
# Load "tidyquant" and "tidyverse" packages library(tidyquant)library(tidyverse)# Get daily stock prices of Apple from the web in a tibble formatApple <-tq_get("AAPL",from="2010-01-04",to="2018-12-31",get="stock.prices")# mutate returns series named as "ret"Ap <- Apple %>%mutate(Date =ymd(date), Beforeclose = dplyr::lag(close),ret =log(close) -log(Beforeclose)) %>%drop_na(ret) #remove "NA"
# Plot log-returns seriesP1 <-ggplot(Ap)+geom_line(aes(x=Date,y=ret),color="gray30")+labs(y="Log Returns", x="") +scale_x_date(date_labels="%Y %b", date_breaks="12 months") +theme_bw()# Plot histogramP2 <-ggplot(Ap)+geom_histogram(aes(ret),binwidth=0.004,col="gray30",fill="gray80")+annotate("text",x=c(-0.1,-0.1),y=c(70,60),label=c("Skewness:-0.1738","Ex.kurtosis:3.5783"),color=c("gray30","gray30"))+labs(y="", x="Log Returns") +theme_bw()# Load "gridExtra" packagelibrary(gridExtra)# Place the two plots on one pagegrid.arrange(P2, P1, nrow=1, top="Apple, Inc. stock price from January 04, 2010 to December 31, 2018")
Correlation matrix: ggcorrplot package
The function ggcorrplot() in the package ggcorrplot can be used to visualize a correlation matrix using ggplot2, which is inspired from the package corrplot.
Consider the motor trend car road tests data set, mtcars. More information about the data can be found by typing ?mtcars
The package plotly can also used to make a contour/and surface plots.
Consider the 87\times 61 altitudes of a volcano, Maunga Whau (Mt Eden) in Auckland, stored in the datasets volcano. More information about the data can be found by typing ?volcano
Consider the same previous data set in the previous slide. Here, we use the package gapminder to compare by continents.
p <-ggplot(gapminder,aes(x=gdpPercap, y=lifeExp, size=pop,color=country)) +geom_point(show.legend=F,alpha=0.7) +scale_color_viridis_d() +scale_size(range=c(2, 12)) +scale_x_log10()+theme_bw() +labs(x="GDP per capita",y="Life expectancy") +facet_wrap(~continent)p +transition_time(year) +labs(title="Year: {frame_time}")
Interactive plots: dygraphs package
The package dygraphs provides R bindings for javascript library dygraphs for time series data.
Consider the three time series giving the monthly deaths from lung diseases in the UK from 1974 to 1979 for both sexes (ldeaths), males (mdeaths) and females (fdeaths).
library(dygraphs)UKLungDeaths <-cbind(ldeaths, mdeaths, fdeaths)dygraph(UKLungDeaths, main="Monthly Deaths from Lung Diseases in the UK")%>%dyOptions(colors=c("#66C2A5","#FC8D62","#8DA0CB"))
Interactive plots: networkD3 package for network graph
The package networkD3 allows the use of interactive network graphs from the D3.js javascript library.
R package crosstalk allows crosstalk enabled plotting libraries to be linked. Through the shared key variable, data points can be manipulated simultaneously on two independent plots.
Q1: (see chapter 3 of Rob Kabacoff book : Data Visualization with R): The Marriage dataset from the package mosaicData contains the marriage records of 98 individuals in Mobile County, Alabama.
Plot the bar chart to display the distribution of wedding participants by race.
plot the distribution of race with modified colors and labels.
plot the distribution of race as percantages.
Sort the bar chart with percent labels.
Q2: (see chapter 4 of Rob Kabacoff book : Data Visualization with R): The mpg dataset from the package ggplot2 is a fuel economy data from 1999 to 2008 for 38 popular models of cars.
Plot stacked bar chart as seen in Section 4.1.1 and add labels to it.
Plot grouped bar chart as seen in Section 4.1.2 and add labels to it.
Q3: (see chapter 11 of Rob Kabacoff book : Data Visualization with R):
Save the graphs, in Q1 and Q2, as several formats (pdf, jpeg, tiff, png, svg, wmf) using the function ggsave().
Save the graphs, in Q1 and Q2, use the menu in plot panel: Plots panel –> Export –> Save as Image or Save as PDF
Source: https://nbisweden.github.io/RaukR-2019/ggplot/presentation/ggplot_presentation.html#1. 
