Msc EDCBA

Marc-Arthur Diaye
Full Professor
University Paris 1 Pantheon-Sorbonne

Data Analytics
Basic notions of R
CLASS 1A
 • http://www.r-project.org/

• CRAN (Comprehensive R Arxiv Network)

• http://cran.r-project.org/

• Rstudio
• https://www.rstudio.com/
 Data
• To find which folder R is currently looking in,
type :
• getwd()
• To change folder (directory) :
• setwd(‘’path’’)
• Important : use / instead of \.
• For instance: setwd(‘’C:/Marc/Master’’)
• Go directly to « Files » or « Session » (if you use
Rstudio)
• From « Files » or « Session », change folder.
 Data
• R can read data sets in text format (ascii)
using the following functions :
• read.table
• scan
• read.fwf
 Data
• R can read also files in Excel, SAS, SPSS,….
• These functions are however not in the
baseline package.
 Data
• Function read.table permits to read a
dataset.
• It is the main function used to read a dataset.
•
• Example: A « txt » file called « coi2006 »
• From this dataset, we can create a dataset
called mydata:
• mydata<-read.table("coi2006.txt" ,
header=TRUE)
 Data
• sep: sep="\t" tells R that the file is tab-
delimited (use " " for space delimited and ","
for comma delimited; use "," for a .csv file).

• row.names: a vector containing the names of

the lines which can be a vector of mode
character, or the number (or the name) of a
variable of the file (by default: 1, 2, 3, ...)
 Data
• col.names: a vector that includes vecteur the
name of the dataset variables (by default :
V1, V2, V3, . . .).

• as.is: controls the conversion of character

variables into factor (if FALSE) or keeps them
in characters (TRUE); as.is can be a logical
vector, numeric or character specifying the
variables preserved in character.
 Data
• Les variantes de read.table sont utiles car
elles ont des valeurs par défaut différentes :
• read.csv(file, header = TRUE, sep = ",",
quote="\"", dec=".", fill = TRUE, ...)
• read.csv2(file, header = TRUE, sep = ";",
quote="\"", dec=",",fill = TRUE, ...)
• read.delim(file, header = TRUE, sep = "\t",
quote="\"", dec=".",fill = TRUE, ...)
• read.delim2(file, header = TRUE, sep = "\t",
quote="\"", dec=",",fill = TRUE, ...)
 Data
• The dataset we will use is already at the .dta
format
• In order to use database at this format, first
download the package called « foreign ».
• You can also directly write in the console :
library("foreign")
• Then duplicate a dataset called « mydata »
from « coi2006.dta ».
• mydata<-read.dta("coi2006.dta")
 Data
• If you use Rstudio, you can directly visualize
the data from the « Environment » part.
• If you use the standard R, you can go to
« Edit », then to « Edit data».

• If you want to have the list of files in the folder that

you use, you can write in the console
• ls()
• Or
• objects()
 Data
• Quit R

• Enter R

• library("foreign")

• mydata<-
read.dta(file="f:/Coi2006/coi2006.dta")
 Data
• To have the list of variables:
• head(mydata)

• To make the columns of the dataset available

for calculus, the data must be attached:
• attach(mydata)
• The variables from the dataset are therefore
available for calculus.
 Data
• You can compactly display the structure of all
variables from the dataset:
• str(mydata)

• You can compactly display the structure of a

specific variable z from the dataset:
• str(mydata$z)
• For instance : str(mydata$sexe)
 Data
• Statistics:

• summary(mydata)
• Provide a summary of all variables from the dataset.

• summary(salnet)
• Provide a summary of variable salnet.

• For numerical variables, summary provide Min, Max,

Q1, Q2, Mean and Q3 (third quartile).

• Missing values are denoted NA.

 Data
• Example of the NET WAGE variable: salnet.

• mean(salnet)
• var(salnet)
• sd(salnet)
• quantile(salnet)
• median(salnet)
• range(salnet)

• boxplot(salnet)
• boxplot(salnet, horizontal=TRUE)
 Data

• Example of the NET WAGE variable: salnet.

• You can first define a list, before performing a

boxplot

• R> b=list("Man"=salnet[sex==1],
"Woman"=salnet[sex==2])
• R> boxplot(b)
 Data

• BOXPLOT.

A boxplot, is a simple diagram that represents the

distribution of a variable.

This diagram is composed of:

- A rectangle that extends from the first to the third
quartile.
- The rectangle is divided by a line corresponding to the
median.
–
 Data

• BOXPLOT.

- This rectangle is completed by two segments of lines.

- To draw them, we first calculate the bounds:
 b− = x1/4− 1.5IQ
 and b+ = x3/4 + 1.5IQ
With IQ the interquartile distance (i.e., the difference
between the 3rd quartile x3/4 and the 1st quartile x1/4).
 Data
• BOXPLOT.
- The smallest and largest observation between
these boundaries is then identified. These
observations are called "adjacent values".
- We draw the line segments linking these
observations to the rectangle.
- Values that are not between adjacent values
are represented by dots and are called
"extreme values".
 Data : More about Boxplot
• boxplot(list(salnet,salbrut))
• By default, whiskers have a maximum length
equal to 1.5 times the size of the box.
• This coefficient can be modified with the range
option.
• You can also change the width of the box with
the width option.
• The names option makes it possible to specify
the labels to be displayed under each box. For
example here one could use names = c("x","y").
 Data : More about Boxplot
• Example:
• boxplot(list(salnet,salbrut),names=c("Net
wage", "Gross wage"))
 Data

• Example of the NET WAGE variable: salnet.

• Compute average net wage by gender:

• R> b=list(sex)
• R> aggregate(salnet,by=b,mean)
• Or :
• R> aggregate(mydata$salnet,by=b,mean)
 Data

• Example of the NET WAGE variable: salnet.

• Mean net wage per gender and PCS (3 :
Managers, 4: Middle managers, 5:
Employees, 6: Blue-collars):
• R> b=list(sex,cscor)
• R> aggregate(salnet,by=b,mean)
• Or :
• R> aggregate(mydata$salnet,by=b,mean)
 Data

• Example of the NET WAGE variable: salnet.

• Standard-deviation of net wage per gender
and PCS (3 : Managers, 4: Middle managers,
5: Employees, 6: Blue-collars):
• R> b=list(sex,cscor)
• R> aggregate(salnet,by=b,sd)
• Or :
• R> aggregate(mydata$salnet,by=b,sd)
 Data

• When a variable includes some missing

values:
• Example of variable r100 (profit).

• R> mean(r100)
• [1] NA
 Data
• When a variable includes some missing
values:
• Example of variable r100 (profit).

• How to deal with it ?

• Answer : create a new database without
missing values:
• R> mydata2<-mydata[!is.na(mydata$r100),]
 Data

• When a variable includes some missing values:

• Example of variable r100 (profit).

• Then compute the mean value of profit using the new

dataset.
• R> mean(mydata2$r100)
• [1] 17605.16

• A simpler solution is to use the option «na.rm »

directly on the original dataset:
• R> mean (r100, na.rm=TRUE)
 Data
• When a variable includes some missing values:
• Suppose that you want to have the number of missing
observations from a a given variable.
• For instance « salnet » and « r100 »
• We know that salnet includes no missing values, while r100
includes some missing values.
• R> sum(is.na(salnet))
• [1] 0
• R> sum(is.na(r100))
• [1] 1072

• You can also use « table »:

• R> table(is.na(r100))
• [1] FALSE TRUE
• 11912 1072
 Data

• Compute the mean of net wage, per gender.

• R> b=list(mydata2$sex)
• R> aggregate(mydata2$r100,by=b,mean)
 Data
• Categorical variable :
• R> table(diplome)
• R> diplome
• 1 10 2 3 4 5 6 7 8 9
• 677 597 903 1864 2961 1066 1228 2350 716 622

• If we want some proportions:

• R> tab<-table(diplome)
• R> prop.table(tab)
• 1 10 2 3 4
• 0.05214110 0.04597967 0.06954713 0.14356131 0.22804991
• 5 6 7 8 9
• 0.08210105 0.09457794 0.18099199 0.05514479 0.04790511
• We can have the same result directly with :
• R> prop.table(table(diplome))
 Data
• Categorical variable:

• Graph of frequencies:
• R> barplot(table(diplome))
 Data
• Categorical variable:

• R> plot(table(diplome))
 4. Data
• Categorical variable:
• Pie:
• pie(table(diplome))
 Data
• Categorical variable :
• Pie:
• pie(table(sex), main="Distribution
Man/Woman", labels=c("Man",
"Woman"),col=c("green", "yellow"))
 Data
• Two categorical variables:
• sex, diplome
• R> xtabs(~sex+diplome)
• Provides a contengency table.
• diplome
• sex 1 10 2 3 4 5 6 7 8 9
• 1 421 479 553 1486 1909 551 785 1271 384 320
• 2 256 118 350 378 1052 515 443 1079 332 302
 Data
• Two categorical variables:

• We can do also:
• R> table(sex,diplome)

• Or:
• R> x<-table(sex,diplome)
• R> x
 Data
• Two categorical variables:
• R> plot(x)
• Provides a bar plot.
 Data
• Two categorical variables:
• R> summary(x)
• Provides the chi-square (Chisq), the number
of degree of freedom (df) and the p-value:
Independence test of two variables (here: sex
and diploma).
 Data
• Two categorical variables:
• R> summary(x)

• Number of cases in table: 12984

• Number of factors: 2
• Test for independence of all factors:
• Chisq = 504.4, df = 9, p-value = 6.532e-103
 Data
• Chi-square Independence Test
• H0 (Null hypothesis): The two distributions are
independent / H1: The two distributions are not
independent.

• summary(x) provides the p-value of the test.

• Fix the probability of the type 1 error (the rejection
of a true null hypothesis)  (1%, 5% or 10%). If the p-
value   then reject H0.
• In our example, p-value = 6.532e-103 <  =1%. Then
we do not reject H0.
 Data
• Two numerical variables:
• Example : salnet (net wage); salbrut (gross
wage)
• R> cor(salnet,salbrut)
• Provides the correlation coefficient (in the
sense of Pearson) betwen the two variables.
• R> cor(salnet,salbrut,method=« spearman »)
provides the correlation coefficient in the
sense of Spearman.
 Data
• Two numerical variables :
• Spearman Correlation
• Spearman compares the order of the values taken by
the two variables.
• Let us assume variables x = (x1,…,xn) and y =
(y1,…,yn).
• The distributions of x and y are ordered in the increasing
direction and renumbered, so that:
• x1<…<xn ; y1<…<yn.
• Spearman correlation coefficient =
• 1- where di = xi-yi
 Data
• Two numerical variables :

• Link between two numerical variables:

• R> plot(salnet~salbrut)
• R> plot(salnet, salbrut)
 Data
• One numerical variable and one categorical
variables:
• Let takes « salnet » and « sex »:
• R> tapply(salnet,sex,mean)
• 1 2
• 26330.01 19942.62
 Data

• R> tapply(salnet,sex,summary)

• $`1`
• Min. 1st Qu. Median Mean 3rd Qu. Max.
• 3210 16930 21010 26330 28490 100000

• $`2`
• Min. 1st Qu. Median Mean 3rd Qu. Max.
• 3444 13580 16930 19940 22780 100000
 Data

• Graph:
• R> b=list("Man"=salnet[sex==1],
"Woman"=salnet[sex==2])
• R> boxplot(b)
 Data
• Remark concerning attach()
• We said that we need to write
• attach(mydata)
• before being able to do any calculus over the variables
of the data set.

• Actually, it is not necessary to go through this step.

But then, you have to specify the dataset that you use.
• Example:
• mean(mydata$salnet)
• table(mydata$sexe)
 Data
• Create a new variable from another:
• We want to create a variable denoted
« lsalnet » defined as the ln of net wage.
• lsalnet<-log(salnet)
• mean(lsalnet)

• We can do also:
• mydata$lsalnet<-log(mydata$salnet)
• mean(mydata$lsalnet)
 Data
• Create a new variable from another:

• Remind that the PCS variable (denoted CSCOR) takes four

modalities : 3 : Manager, 4: Middle Manager, 5: Employee,
6: (Blue-collar) Worker
• From CSCOR, we want to crate a new variable called PCS2.
With:
• PCS2 = TRUE if cscor = 3 ou 4
• PCS2 = FALSE otherwise
• pcs2<-cscor %in% c(3,4)

• On peut faire la même chose avec

• pcs2<-(cscor %in% c(3,4))
 Data
• Create a new variable from another:
• If we want to create a binary variable 0/1:
CSP3

• pcs3<-rep(0,length(cscor))
• pcs3[cscor %in% c(3,4)]=1
 Data
• Create a new variable from another:
• Practice.

• Create a binary variable from the « sex »

variable.
• Remind: sex=1 if man; 2 if woman
 Data

• man=rep(0,length(sex))
• table(man)

• man[sex=="1"]=1
• table(man)
 Data

• Practice : Impliq variable

• Create from this variable a binary effort
variable.

• Answer:
• Effort (impliq)
• effort<-rep(0,length(impliq))
• effort[impliq =="3"]=1
 Data
• More concerning the Plot function
• xlim, ylim : set the lower and upper limits of
the axes (two-element vectors).
• xlab, ylab : allow you to specify the axis
legend (character mode).
• main : allows to put a title above the graph
(character mode).
• pch : defines the symbol representing the
points; an integer from 1 to 25, or any
character in quotation marks.
 Data
• More concerning the Plot function
• col: specifies the color of symbols
("blue","red" etc. the exhaustive list of
available colours can be get with colors()),

• bty: controls the shape of the frame; default

square ("o"), L ("l"), U ("u"), C ("c"), 7 ("7") or
square brackets ("]").
 Data
• More concerning the Plot function

• We can play on the size of the symbols thanks to

the option cex.

• By default cex=1 ; however we can provide to

the software, a positive number that represents
a multiplicative coefficient relative to the default
size (a value between 0 and 1 to reduce the size,
or greater than 1 to increase it).
 Data
• More concerning the Plot function

• In the same way the options cex.axis, cex.lab

and cex.main control the size of the graduations
of the axes, labels of the axes and the title.

• To change the style of the text, we use the font

option, which also comes in the forms font.axis,
font.lab and font.main, (1 for normal, 2 for italic,
3 for bold and 4 for bold italic) .
 Data
• More concerning the Hist function

• We saw that to draw a histogram, the basic

command is the hist function :
• hist(x)

• Some options of the hist function:

• breaks : allows you to specify the break points
between the bars of the histogram, either as a vector
or as a number of bars.
• freq : allows to choose the frequency (freq=TRUE,
default option), or the proportion (freq=FALSE).
 Data
• More concerning the Hist function

• Some options of the hist function :

• col : indicates the color to fill the bars.
• plot : if plot=FALSE, the histogram is not drawn
and the function returns the list of break points
and numbers.
• right : allows to choose intervals of type ]a, b] if
right=TRUE (by default, [a, b[).
 Data
• More concerning the Hist function

• Example:
• hist(salnet, freq = FALSE, main = "Histogram
NET Wage")
 Data
• More concerning the Hist function

• Example:
• hist(salnet, freq = TRUE, main = "Histogram
Net Wage")
 Data
• More concerning the Hist function

• Example:

• hist(salnet, freq = TRUE, col = "blue", main =

"Histogram Net Wage")
 Data
• More concerning the Hist function

• Example: Add some cuts

• Suppose that you want two cuts:

• hist(salnet, breaks = 2, freq = FALSE, col =

"blue", main = "Histogram Net Wage")
• hist(salnet, breaks = 2, freq = TRUE, col =
"blue", main = "Histogram Net Wage")
 Data
• More concerning the Hist function

• Transform a variable inside the hist function:

• hist(log(salnet), freq = FALSE, main =

"Histogram Net Wage")
 Data
• More concerning the Hist function

• Transform a variable inside the hist function +

smooth:

• lines(density(log(salnet)), col= "red")

 Data
#Remove some variables from the dataset : example with
cscor, age and effl_corr
myvars <- names(mydata) %in% c("cscor", "age", "effl_corr")
newdata <- mydata[!myvars]

#We can check that the variable cscor no more exists in

newdata
table(newdata$cscor)
< table of extent 0 >

#Remove the 3rd and 5th variables

newdata <- mydata[c(-3,-5)]
 Data

#Remove variables TYPEMPLOI and IMPLIQ in the

original database mydata
mydata$typemploi <- mydata$impliq <- NULL

We can do also
mydata$typemploi <- NULL
mydata$impliq <- NULL

#If you don’t need to specify the dataset, then

typemploi <- impliq <- NULL
 Data
# Select some variables: Example of cscor, age,
effl_corr
myvars2 <- c("cscor", "age", "effl_corr")
newdata <- mydata[myvars2]
fix(newdata)

# Select the 1st variable and variables 5 to 10

newdata <- mydata[c(1,5:10)]
 Data
# Select the 20 first observations of the dataset
newdata <- mydata[1:20,]
fix(newdata)

#Select observations that fulfill some conditions

#Example: Select executive woman under 40 years old
newdata <- mydata[ which(mydata$sex=="2" &
mydata$cscor=="3" & mydata$age <= 40), ]

# or if we have first performed attach(mydata)

newdata <- mydata[ which(sex=="2" & cscor=="3" & age
<= 40), ]
 Data
The best way to select observations is to use
the subset function

Syntax: subset( )
 Data
Example:
• Select executive women less than 40 years old.
• Keep variables : sex, cscor, age, salnet, siren,
depnaiss, effl_corr, couple.

#Use function subset

newdata <- subset(mydata, sex=="2" & cscor=="3"
& age <= 40, select=c(sex,cscor,age,salnet,siren,
depnaiss,effl_corr,couple))
 Data
Practice 1:
• Create from COI2006 two datasets including
respectively:
• Stressed men less than 35.
• Stressed men at least 35.
• Compare the average net wage of the two
groups.
 Data
Practice 2:
• COI2006 includes two gender variables : sexe
and sex. However these variables come from
two different statistical sources. As a
consequence, they do not agree all the times.
• Compare both variables.
• Is it possible to have a convincing answer about
the agents’ true gender, when the two variables
disagree ?
Introduction to GGPLOT 2
CLASS 1 B
 Comparing densities between groups
• library("foreign")
• mydata=read.dta(file="F:/COI2006/coi2006.dta")
• attach(mydata)

• library (lattice)
• densityplot(~salnet|sex)

#If you need to specify the name of the data set

• density(~salnet|sex, data=mydata)
 Practice 1
Comparing (Kernel) densities between group

• In the same graph, draw the densities of net

wage (salnet variable) for managers and non
managers, for women and for men.
• Reproduce the following graph.
 Solution 1
• library(lattice)

• v_manager<-rep("Non manager",length(cscor))
• v_manager[cscor %in% c(3,4)]="Manager"
• v_homme<-rep("Woman",length(sex))
• v_homme[sex=="1"]="Man"

• densityplot(~salnet|v_homme,
groups=v_manager, data=mydata,
auto.key=list(space="right"), main="Density Net
Wage Manager/Non manager, Woman/Man")
 Data Visualization
• Visualisation is a fundamentally human
activity.

• A good visualization may show us things that

we did not expect, or raise new questions
about the data.
 Datavisualization
• The majority of dataviz packages are part of the
so-called tidyverse.

• The packages in the tidyverse share a common

philosophy of data and R programming, and are
designed to work together naturally.

• We can install the complete tidyverse with a

single line of code:

• install.packages("tidyverse")
 Datavisualization
• Then :
• library(tidyverse)

• You will see :

• -- Attaching packages --------------------------------------- tidyverse 1.2.1

• v ggplot2 3.1.0 v purrr 0.3.2
• v tibble 2.1.1 v dplyr 0.8.0.1
• v tidyr 0.8.3 v stringr 1.4.0
• v readr 1.3.1 v forcats 0.4.0
• -- Conflicts ------------------------------------------ tidyverse_conflicts() --
• x dplyr::filter() masks stats::filter()
• x dplyr::lag() masks stats::lag()
 Datavisualization
• This tells us that tidyverse is loading the ggplot2,
tibble, tidyr, readr, purrr, and dplyr packages.

• These are considered to be the core of the tidyverse

because we will use them in almost every analysis.

• Packages in the tidyverse change fairly frequently.

• We can see if updates are available, and optionally

install them, by running :
• tidyverse_update()
 Datavisualization
• Creating a ggplot

• With ggplot2, we begin a plot with the function

ggplot()

• ggplot() creates a coordinate system that we can add

layers to.

• The first argument of ggplot() is the dataset to use in

the graph.
• So ggplot(data = mydata) creates an empty graph.
 Datavisualization
• Creating a ggplot

• We complete our graph by adding one or more

layers to ggplot().

• For instance, the function geom_point() adds a

layer of points to our plot, which creates a
scatterplot.

• ggplot2 comes with many geom functions that

each add a different type of layer to a plot.
 Datavisualization
• Creating a ggplot

• Each geom function in ggplot2 takes a mapping

argument. This defines how variables in our dataset
are mapped to visual properties.

• The mapping argument is always paired with aes(),

and the x and y arguments of aes() specify which
variables to map to the x and y axes.

• ggplot2 looks for the mapped variables in the data

argument.
 Datavisualization
• Creating a ggplot

• Example :
• ggplot(data=mydata)+geom_point(mapping =
aes(x = salnet, y = age))
 Datavisualization
• Creating a ggplot

• We can convey information about our data by

mapping the aesthetics in our plot to the variables in
our dataset.

• For example, we can map the colors of our points to

the “sex” variable to reveal the “sex” of each .

• Example :
• ggplot(data=mydata)+geom_point(mapping = aes(x =
salnet, y = age, color=sex))
 Datavisualization
• Creating a ggplot

• Example with shape :

• ggplot(data=mydata)+geom_point(mapping =
aes(x = salnet, y = age, shape=sex))
 Datavisualization
• Creating a ggplot

• Example with alpha :

• ggplot(data=mydata)+geom_point(mapping =
aes(x = salnet, y = age, alpha=sex))
 Datavisualization
• Creating a ggplot

• We can also set the aesthetic properties of

our geom manually.
• For example, we can make all of the points in
our plot blue:
• ggplot(data=mydata)+geom_point(mapping =
aes(x = salnet, y = age), color="blue")
 Datavisualization
• Creating a ggplot

• The color does not convey information about a

variable, but only changes the appearance of the plot.

• To set an aesthetic manually, set the aesthetic by

name as an argument of our geom function;

• We will need to pick a level that makes sense for that

aesthetic:
• The name of a color as a character string;
• The size of a point in mm;
• The shape of a point as a number (see the below figure).
 Datavisualization
• Creating a ggplot

• Example:
• ggplot(data=mydata)+geom_point(mapping = aes(x =
salnet, y = age), shape=23)

• ggplot(data=mydata)+geom_point(mapping = aes(x =
salnet, y = age), color="blue", shape=11)

• ggplot(data=mydata)+geom_point(mapping = aes(x =
salnet, y = age), color="blue", shape=19, size=3)
SHAPE=23
COLOR=BLUE , SHAPE=11
COLOR=BLUE , SHAPE=19
 Datavisualization
• Creating a ggplot

• Example: Add a title using ggtitle

• ggplot(data=mydata)+geom_point(mapping =
aes(x = salnet, y = age), color="blue",
shape=11) + ggtitle("Shalom to my niece
Rivka")
 Datavisualization
• Creating a ggplot

• Example: Add a title using labs

• ggplot(data=mydata)+geom_point(mapping =
aes(x = salnet, y = age), color="blue",
shape=11) + labs(title="Shalom to my niece
Rivka")
 Datavisualization
• Creating a ggplot

• Example: Change the name of the axis using

xlab and ylab

• ggplot(data=mydata)+geom_point(mapping =
aes(x = salnet, y = age), color="blue",
shape=11) + ggtitle("Shalom to my niece
Rivka") + xlab("Net wage") + ylab("Ageeeee")
 Datavisualization
• Creating a ggplot

• Example: We can have the same result (i.e.,

change the name of the axis) using labs

• ggplot(data=mydata)+geom_point(mapping =
aes(x = salnet, y = age), color="blue",
shape=11) + labs(title="Shalom to my niece
Rivka", x="Net wage", y="Ageeeee")
 Datavisualization
• Creating a ggplot

• Example: Change the color, size and type of the main

title and the axis titles, using theme

• ggplot(data=mydata)+geom_point(mapping = aes(x =
salnet, y = age), color="blue", shape=11) +
labs(title="Shalom to my niece Rivka", x="Net wage",
y="Ageeeee")+theme(plot.title =
element_text(color="red", size=14, face="bold.italic"),
• axis.title.x = element_text(color="blue", size=14,
face="bold"), axis.title.y =
element_text(color="#993333", size=14, face="bold"))
 Hexadecimal color code chart
• Colors can specified as a hexadecimal RGB (Red, Green,
Blue) triplet, such as "#0066CC".

• The first two digits : are the level of red,

• the next two digits : green,
• and the last two digits : blue.

• The value for each ranges from 00 to FF in hexadecimal

(base-16) notation, which is equivalent to 0 and 255 in
base-10.

• For example, in the table below (slide 30), “#FFFFFF” is

white and “#990000” is a deep red.
http://www.visibone.com
 Datavisualization
• Creating a ggplot

• Example: Remove main title or axis titles

using theme

• theme(plot.title = element_blank(),
• axis.title.x = element_blank(), axis.title.y =
element_blank())
 Datavisualization
• Creating a ggplot

• One way to add additional variables is with

aesthetics.

• Another way, particularly useful for categorical

variables, is to split our plot into facets, subplots
that each display one subset of the data.

• To facet our plot by a single variable, we can use

facet_wrap().
 Datavisualization
• Creating a ggplot

• The first argument of facet_wrap() should be

a formula, which we create with ~ followed
by a variable name.

• The variable that we pass to facet_wrap()

should be discrete.
 Datavisualization
• Creating a ggplot

• Example :

• ggplot(data = mydata) + geom_point(mapping =

aes(x = age, y = salnet)) + facet_wrap(~sex, nrow
= 2)

• ggplot(data = mydata) + geom_point(mapping =

aes(x = age, y = salnet)) + facet_wrap(~cscor,
nrow = 2)
 Datavisualization
• Creating a ggplot

• To facet our plot on the combination of two

variables, we can add facet_grid() to our plot
call.

• The first argument of facet_grid() is also a

formula.

• This time the formula should contain two

variable names separated by a ~.
 Datavisualization
• Creating a ggplot

• Example :
• ggplot(data = mydata) +
geom_point(mapping = aes(x = age, y =
salnet)) + facet_grid(sex ~ cscor)
 Datavisualization
• Creating a ggplot

• If we prefer to not facet in the rows or columns

dimension, we can use a “.” instead of a variable
name.
• Example :
• ggplot(data = mydata) + geom_point(mapping =
aes(x = age, y = salnet)) + facet_grid(. ~ cscor)
• ggplot(data = mydata) + geom_point(mapping =
aes(x = age, y = salnet)) + facet_grid(sex ~ .)
 Datavisualization
• Creating a ggplot : geometric objects

• Since then, we have used geom_point.

• However for bar charts we can use bar geoms;
• For line charts, we can use line geoms;
• For boxplots, we can use boxplot geoms;
• (…)
• We can use the smooth geom, in order to have a smooth
line fitted to the data.
• the data is fitted using the so-called Loess method; Loess short
for Local Regression is a non-parametric approach that fits
multiple regressions in local neighborhood;
• Or the so-called Gam method; Gam short for generalized
additive model.
 Datavisualization
• Creating a ggplot : geometric objects

• To change the geom in our plot, we have

simply to change the geom function that we
add to ggplot().

• Example: Smooth the plot

• ggplot(data = mydata) +
geom_smooth(mapping = aes(x = age, y =
salnet))
 Datavisualization
• Creating a ggplot : geometric objects

• Every geom function in ggplot2 takes a mapping argument.

• However, not every aesthetic works with every geom.

• We cant set the shape of a point, but we cannot set the “shape” of
a line.

• On the other hand, we can set the linetype of a line.

• geom_smooth() will draw a different line, with a different linetype,

for each unique value of the variable that we map to linetype.
 Datavisualization
• Creating a ggplot : geometric objects

• Example:
• ggplot(data = mydata) +
geom_smooth(mapping = aes(x = age, y =
salnet, linetype = sex))
 Datavisualization
• Creating a ggplot : geometric objects

• geom_smooth(), uses a single geometric

object to display multiple rows of data.
• Therefore we can set the group aesthetic to a
categorical variable to draw multiple objects.
• ggplot2 will draw a separate object for each
unique value of the grouping variable.
 Datavisualization
• Creating a ggplot : geometric objects

• In practice, ggplot2 will automatically group the

data for these geoms whenever we map an
aesthetic to a discrete variable (as in the linetype
example).

• It is convenient to rely on this feature because

the group aesthetic by itself does not add a
legend or distinguishing features to the geoms.
 Datavisualization
• Creating a ggplot : geometric objects

• Example:
• ggplot(data = mydata) + geom_smooth(mapping =
aes(x = age, y = salnet, group = sex))

• ggplot(data = mydata) + geom_smooth(mapping =

aes(x = age, y = salnet, color = sex))

• ggplot(data = mydata) + geom_smooth(mapping =

aes(x = age, y = salnet, color = sex), show.legend =
FALSE)
 Datavisualization
• Creating a ggplot : geometric objects

• To display multiple geoms in the same plot, we

can add multiple geom functions to ggplot().

• Example :
• ggplot(data = mydata) + geom_point(mapping =
aes(x = age, y = salnet)) +
geom_smooth(mapping = aes(x = age, y =
salnet))
 Datavisualization
• Creating a ggplot : geometric objects

• This, however, introduces some duplication in

our code.

• Imagine if we wanted to change the y-axis to

display salbrut instead of salnet.

• We will need to change the variable in two

places, and we may forget to update one.
 Datavisualization
• Creating a ggplot : geometric objects

• We can avoid this kind of repetition by

passing a set of mappings to ggplot().

• ggplot2 will treat these mappings as global

mappings that apply to each geom in the
graph.
 Datavisualization
• Creating a ggplot : geometric objects

• Example :
• ggplot(data = mydata, mapping = aes(x = age,
y = salnet)) + geom_point() + geom_smooth()
 Datavisualization
• Creating a ggplot : geometric objects

• If we place mappings in a geom function, ggplot2

will treat them as local mappings for the layer.
• It will use these mappings to extend or overwrite
the global mappings for that layer only.

• This makes it possible to display different

aesthetics in different layers.
 Datavisualization
• Creating a ggplot : geometric objects

• Example:
• ggplot(data = mydata, mapping = aes(x = age,
y = salnet)) + geom_point(mapping =
aes(color = sex)) + geom_smooth()
 Datavisualization
• Creating a ggplot : geometric objects

• Example:
• ggplot(data = mydata, mapping = aes(x = age,
y = salnet)) + geom_point(mapping =
aes(color = sex)) + geom_smooth(data =
filter(mydata, sex == "1"))
• #Here the smooth line is the one for men
 Datavisualization
• Creating a ggplot : geometric objects

• Example:
• ggplot(data = mydata, mapping = aes(x = age,
y = salnet)) + geom_point(mapping =
aes(color = sex)) + geom_smooth(data =
filter(mydata, sex == “2"))
• #Here the smooth line is the one for women
 Datavisualization
• Creating a ggplot : geometric objects

• Example:
• ggplot(data = mydata, mapping = aes(x = age,
y = salnet)) + geom_point(mapping =
aes(color = sex)) + geom_smooth(data =
filter(mydata, sex == “2"), se=FALSE)
• #Here the smooth line is the one for women
and we have removed the standard-error
representation
 Datavisualization
• Creating a ggplot : geometric objects

• Practice 2

• Recreate the code necessary to have the

following graph:
Practice 2
 Datavisualization
• Creating a ggplot : geometric objects

• A solution to Pratice 2:
• ggplot(data = mydata, mapping = aes(x = age,
y = salnet, color=sex)) + geom_point() +
geom_smooth()