Fundamentals of Data Science for EESS

R sessions

Introduction

• Who has used R before ?

• What other programming language have you used before ?

• For those who are experts in R

  • please refrain to answer during this session...
  • help your neighbor...

• Two special slide formatting

Your turn...

Warning

Introduction

Computer languages

Introduction

History of R

• Mid 1970s - S Language for Statistical Computing conceived by John Chambers, Rick Becker, Trevor Hastie, Allan Wilks and others at Bell Labs

• Early 1990's - R was first implemented in the early 1990’s by Robert Gentleman and Ross Ihaka, both faculty members at the University of Auckland.

• 1995 - Open Source Project

• 1997 - Managed by the R Core Group

• 2000 - First release of R

• 2011 - First release of R studio

• Historical notes - Paper from 1998

Introduction

Why use R ?

• Script vs. Menu driven software (e.g. Excel)
  • Can be re-rerun with new data
  • Reproducible workflow
• Open source
  • Huge number of libraries
  • Tidy "universe" : tidyverse and ggplot2
    • Very easy to manipulate tables (select columns, create new variables)
    • High quality graphics
• Work environment
  • R studio
• Document your data processing
  • R markdown
  • Create HTML, pdf, presentations
• Share your data and workflow
  • GitHub

Introduction

What can you do with R ?

Resources

Books and Manuals

• Applied Statistics with R : Quite simple introduction with emphasis on Stats
• R intro : Very good introduction to R, short and clear
• R in a nutshell : Many many receipes to solve all your questions
• R graphics cook book : very good for graphics

Resources

On line courses and web sites

• Coursera
• Pluralsight - Not free
• Quick-R, very simple

Resources

Cheat sheets

• R basics
• ggplot2
• dplyr

Resources

Forum

• https://stackoverflow.com/
• http://r-statistics.co/
• https://stats.stackexchange.com/
• https://www.r-bloggers.com/

Let's get started

Setup

• Install R

• Install R studio

Let's get started

The R studio interface

Let's get started

Create a new project

• Open R studio
• Create new project for the course in a new directory
  • e.g. Experimental design course

Let's get started

Your first script

print("Hello world")

[1] "Hello world"

Two ways to proceed

1. Type directly in command window

2. Create a new script

Type in script window, select and execute (CTRL-R)

The R language

Everything in R is an object

• Assignement done with <-

> x <- 1
> y <- 2
> x + y

[1] 3

> z <- x + y
> z

[1] 3

The R language

= can be used instead of <- but refrain from it (not good style)

> z = x + y

You can view the values of the objects in R-studio environment window (top-right)

The R language

R is case sensitive

> Z

> Z

Error in eval(expr, envir, enclos): objet 'Z' introuvable

The R language

Rules for naming objects

• Use
  • letters
  • numbers
  • the dot
  • the underscore (not the minus sign !)

• Start always with a letter
  • Myvariable, Myvariable1, Myvariable.1,Myvariable-01 are OK
  • 1Myvariable, My-variable, Myvariable@ are not OK

The R language

Use consistent naming

Five conventions

• alllowercase: e.g. adjustcolor
• period.separated: e.g. plot.new
• underscore_separated: e.g. numeric_version
• lowerCamelCase: e.g. addTaskCallback
• UpperCamelCase: e.g. SignatureMethod

Prefer third one, much more easy to read

• Use names for objects : last_name
• Use verbs for function : build_name
• Think about best order
  • e.g. prefer maybe name_last because then you can have name_first, name_full...
  • and you identify that all these objects are related to a name...

R objects

Data types

• character: "Daniel", "This is a course in R", 'Donald'

• numeric: 2, 15.5, 10e-3

• integer: 2L (the L tells R to store this as an integer)

• date: 2018-02-25

• logical: TRUE, FALSE

• complex: 1+4i (complex numbers with real and imaginary parts)

• No data "NA"

• Not a number "NaN" (e.g. division by zero)

R objects

Data structures

• Vector

• List

• Matrix

• Data frames

• Function

Vectors

The basic R structure is a vector: $$\begin{bmatrix}10 \\ 20 \\ 30 \end{bmatrix}$$

A vector can contain only a single element $$\begin{bmatrix}10 \end{bmatrix}$$

Assign a value to a vector

x <- 10
x

[1] 10

Vectors

Assign several elements

x <- c(10, 20, 30)
x

[1] 10 20 30

Assign range

x <- 10:30
x

 [1] 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Vectors

Assign characters

PoTU <- c("Donald", "Trump")
PoTU

[1] "Donald" "Trump"

Assign logical

flags <- c(TRUE, FALSE, TRUE)
flags

[1]  TRUE FALSE  TRUE

Vectors

Access specific elements of a vector

First

x[1]

[1] 10

Range

x[1:5]

[1] 10 11 12 13 14

Remove one element

x[-1]

 [1] 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Vectors

Determine object properties

Apply functions (we will come back to functions latter)

• typeof() - what is the object’s data type (low-level)?
• length() - how long is it? What about two dimensional objects?

typeof(x)
length(x)

[1] "integer"

[1] 21

Operators

Arithmetic Operators

Operators

Arithmetic Operators

We are performing vector operations !

$$\begin{bmatrix} 1\\2\\3\\..\end{bmatrix}+\begin{bmatrix}1\\2\\3\\..\end{bmatrix}=\begin{bmatrix}2\\4\\6\\..\end{bmatrix}$$

Operators

Arithmetic Operators

Vector one element

 x <- 1
 y <- 2
 z <- x + y
 z

[1] 3

Operators

Arithmetic Operators

Vector several elements

# Two instructions on the same line
 x <- 1:9;  y <- 1:9
 z <- x + y
 z

[1]  2  4  6  8 10 12 14 16 18

Operators

Arithmetic Operators

What happens when the vectors have different number of elements ?

x <- 1:9
y <- 1
z <- x + y
z

[1]  2  3  4  5  6  7  8  9 10

Equivalent to

y <- c(1, 1, 1, 1, 1, 1, 1, 1, 1)

The recycling rule...

Operators

Can we add logical ?

x <- TRUE
y <- FALSE
z <- x + y
z

[1] 1

Operators

Can we add logical ?

No error but...

The resulting variable is transformed to a numeric

typeof(x)

[1] "logical"

typeof(z)

[1] "integer"

Operators

Logical Operators

Operators

Logical Operators

x <- TRUE
y <- FALSE
z1 <- x | y
z2 <- x == y

[1] TRUE

[1] FALSE

Operators

Can we add characters ?

first <- "Donald"
last <- "Trump"
full <- first + last

Generates an error

Error in first + last: argument non numérique pour un opérateur binaire

Functions

Functions perform specific task on objects

• e.g. to concatanate strings we use paste()

paste(first, last)

[1] "Donald Trump"

• Functions take arguments and return an object called result

• To know the arguments use ?

  ? paste()  # Do not forget the parenthesis

• Can go directly to Help panel and type function name

Functions

Help

Functions

Help

Functions

Getting what you want

Let's apply paste :

paste(first, last)

[1] "Donald Trump"

paste(first, last, sep = "_")

[1] "Donald_Trump"

Functions

Write your own function

If you write 3 times the same piece of code write a function...

my_sum <- function(first_number, second_number) {
    c <- first_number + second_number
    return(c)
}

• my_sum : function name
• first_number, second_number : arguments
• instructions are enclosed by braces ({})
• return() : the value(s) returned

More compact way

 my_sum <- function(first_number, second_number) {first_number + second_number}

Functions

Call your function

my_sum(10, 20)

[1] 30

• better

my_sum(first_number = 10, second_number = 20)

[1] 30

Functions

Examples of functions

Most of the time you do not have to write functions because someone has already written one for what you want to do...

• Sum

x <- 1:100
sum(x)

[1] 5050

• Normal distribution

y <- rnorm(10, mean = 0, sd = 1)
y

 [1]  1.6731915 -0.2498182  1.0774267  0.7086024 -0.3065056  2.2168636
 [7]  0.7236404  0.2397608 -0.1206248 -1.2904100

Functions

Statistics

mean(y)

[1] 0.4672127

sd(y)

[1] 1.033406

Sample more points... 10,000 instead of 100

y <- rnorm(10000, mean = 0, sd = 1)
mean(y)

[1] -0.005369128

sd(y)

[1] 0.9956479

Functions

Plot

library(graphics)
hist(y)

Packages

Packages are set of functions that have a common goal

They are really the strength of R

And these are only the "official"" packages. You can find more on GitHub

Packages

Installing a package

Download on your computer the package you need

Packages

Using a package

To use functions from the package

• use the syntax package::function

stringr::str_c(first, last, sep = " ")

[1] "Donald Trump"

• load the package with the library function

library(stringr)
str_c(first, last, sep = " ")

[1] "Donald Trump"

Sometimes functions from different libraries have similar names

Packages

List installed packages

Other objects

• List

• Matrix

• Factors

• Data frames

Data frames

What is it ?

• Table mixing different types of columns (an Excel table...)
• However within a column all values are similar, e.g. numeric, logical, character

  df <- data.frame(label = letters[1:6], 
                   id = 1:6, 
                   value = rnorm(6, mean = 0, sd = 1), 
                   flag=c(TRUE, FALSE),                   # recycling rule
                   stringsAsFactors = FALSE)
  df

  label id      value  flag
1     a  1 -0.5324537  TRUE
2     b  2 -0.2858342 FALSE
3     c  3 -0.7311013  TRUE
4     d  4 -1.2440367 FALSE
5     e  5 -0.8671309  TRUE
6     f  6 -2.4274570 FALSE

Data frames

Useful functions

dim(df)  # returns the dimensions of data frame

[1] 6 4

nrow(df)  # number of rows

[1] 6

ncol(df)  # number of columns

[1] 4

Data frames

Useful functions

str(df)  # structure of data frame - name, type and preview of data in each column

'data.frame':    6 obs. of  4 variables:
 $ label: chr  "a" "b" "c" "d" ...
 $ id   : int  1 2 3 4 5 6
 $ value: num  -0.532 -0.286 -0.731 -1.244 -0.867 ...
 $ flag : logi  TRUE FALSE TRUE FALSE TRUE FALSE

colnames(df)  # columns names

[1] "label" "id"    "value" "flag"

Data frames

Access specific value

• Use the df[i,j] notation, first index corresponds to row, second index to column

df[5, 3]

[1] -0.8671309

• Specifiy the name of the column

df[5, "value"]

[1] -0.8671309

Data frames

Access specific column

• Use the df[i,j] notation

df[, 3]

[1] -0.5324537 -0.2858342 -0.7311013 -1.2440367 -0.8671309 -2.4274570

df[, "value"]

[1] -0.5324537 -0.2858342 -0.7311013 -1.2440367 -0.8671309 -2.4274570

Data frames

Access specific column

• Use $notation

df$value

[1] -0.5324537 -0.2858342 -0.7311013 -1.2440367 -0.8671309 -2.4274570

This can be used to access a specific value

• $ for the column, [i] for the row

df$value[5]

[1] -0.8671309

Data frames

Access row

• Use the df[i,j] notation

df[1, ]

  label id      value flag
1     a  1 -0.5324537 TRUE

Data frames

Access specific rows

• Rows for which the value of id <= 3

   df[df$id <= 3,]

  label id      value  flag
1     a  1 -0.5324537  TRUE
2     b  2 -0.2858342 FALSE
3     c  3 -0.7311013  TRUE

Select lines for which the label is c

df[df$label == "c", ]

  label id      value flag
3     c  3 -0.7311013 TRUE

This syntax is complicated - tidyverse packages make it much more easy to manipulate and remember

Next class: 02 - Markdown

What you will learn :

• Mix text, R code and R output in a single document
• Produce documents as HTML, pdf or even Word from the same template

Next class: 03 - Git

Resources

• Pro GIT: https://git-scm.com/book/en/v2
• GitHub guide: https://guides.github.com
• Happy Git with R: https://happygitwithr.com/

Software

• Git: https://git-scm.com
• Git desktop: https://github.com
• GitHub account: https://desktop.github.com