source: Andrie de Vries

David Robinson summarized the workflow on his laptop

Whenever you’re learning a new tool, for a long time you’re going to suck…
But the good news is that it's typical, that’s something that happens to everyone, and it’s only temporary.
– Hadley Wickham

all elements of an atomic vector must be the same type source

concatenate elements of same type

x <- c(1, 1.24, "6")
x
## [1] "1"    "1.24" "6"
is.vector(x)
## [1] TRUE
x[1] # access 1st element
## [1] "1"

See if we enter 6 as character

x <- c(1, 1.24, "6")
x
## [1] "1"    "1.24" "6"
is.vector(x)
## [1] TRUE
is.list(x)
## [1] FALSE
is.atomic(x)
## [1] TRUE

Lists are objects that could contain anything

l <- list(a = 1:3, b = c("hello", "bye"), data = head(iris, 2))
is.vector(l)

## [1] TRUE

is.list(l)

## [1] TRUE

is.atomic(l)

## [1] FALSE

l[1]

## $a
## [1] 1 2 3

l[[1]]

## [1] 1 2 3

# str(l)

vector

x
## [1] "1"    "1.24" "6"
x[1]
## [1] "1"
x[-2] # return 1st and 3rd elements
## [1] "1" "6"
x[c(1, 3)] # return 1st and 3rd elements
## [1] "1" "6"

list

l[2]      # return a list
## $b
## [1] "hello" "bye"
l[2][1]   # makes no sense
## $b
## [1] "hello" "bye"
l[[2]]    # return a vector
## [1] "hello" "bye"
l[[2]][1] # return 1st element, 2nd atomic vector
## [1] "hello"

f <- rep(c("Male", "Female"), each = 4)
f
## [1] "Male"   "Male"   "Male"   "Male"   "Female" "Female" "Female" "Female"
ff <- factor(f)
ff # levels are like dictionnary keys
## [1] Male   Male   Male   Male   Female Female Female Female
## Levels: Female Male
str(ff) # structure of factor, integer behind
##  Factor w/ 2 levels "Female","Male": 2 2 2 2 1 1 1 1

table(ff) # useful command to count entities
## ff
## Female   Male 
##      4      4
ff[3] <- "t" # error! "t" not in levels
## Warning in `[<-.factor`(`*tmp*`, 3, value = "t"): invalid factor level, NA
## generated

matrix

matrix(data = 1:6, nrow = 2, ncol = 3)

##      [,1] [,2] [,3]
## [1,]    1    3    5
## [2,]    2    4    6

which are lists where all columns have equal length

class(iris)
## [1] "data.frame"
class(unclass(iris))
## [1] "list"

Actually, better to use tbl_df tibble diff rstudio blog

data_frame() does much less than data.frame():

• never changes the type of the inputs (e.g. it never converts strings to factors!)
• never changes the names of variables
• never creates row.names()
• never print ALL rows

but display

• column types
• groups (dplyr::group_by())

data frame are lists

mtcars[[1]]

##  [1] 21.0 21.0 22.8 21.4 18.7 18.1 14.3 24.4 22.8 19.2 17.8 16.4 17.3 15.2
## [15] 10.4 10.4 14.7 32.4 30.4 33.9 21.5 15.5 15.2 13.3 19.2 27.3 26.0 30.4
## [29] 15.8 19.7 15.0 21.4

mtcars[["mpg"]]

##  [1] 21.0 21.0 22.8 21.4 18.7 18.1 14.3 24.4 22.8 19.2 17.8 16.4 17.3 15.2
## [15] 10.4 10.4 14.7 32.4 30.4 33.9 21.5 15.5 15.2 13.3 19.2 27.3 26.0 30.4
## [29] 15.8 19.7 15.0 21.4

but $ is a shorthand for [[

mtcars$mpg

##  [1] 21.0 21.0 22.8 21.4 18.7 18.1 14.3 24.4 22.8 19.2 17.8 16.4 17.3 15.2
## [15] 10.4 10.4 14.7 32.4 30.4 33.9 21.5 15.5 15.2 13.3 19.2 27.3 26.0 30.4
## [29] 15.8 19.7 15.0 21.4

• Package management (including building)
• Console to run R, with syntax highligther
• Editor to work with scripts / markdown
• auto-completion using TAB
• Cheatsheets
• Keyboard shortcuts

Cmd + Enter (mac) orCtrl + Enter (PC): sends the line or selection from the editor to the console and runs it.

↑: in the console browse previous commands

Recommended in the r4ds To get a clean environment at start-up

Solve most issues with working directories, get rid of setwd()

magrittr by Stefan Milton Bache

set.seed(124)
x <- rnorm(10)
mean(x)

## [1] 0.2147669

round(mean(x), 3)

## [1] 0.215

set.seed(124)
rnorm(10) %>% mean %>% round(3)

## [1] 0.215

natural from left to right.

Even better with one instruction per line and indentation

set.seed(124)
rnorm(10) %>%
  mean %>%
  round(3)

## [1] 0.215

x %>% f(.) 
#stands for (pipe to the 1st function argument)
f(x)

When not 1st argument

sub("b", "B", letters[1:3])

## [1] "a" "B" "c"

letters[1:3] %>% sub("b", "B") # raises warning!

## Warning in sub(., "b", "B"): argument 'pattern' has length > 1 and only the
## first element will be used

## [1] "B"

letters[1:3] %>% sub("b", "B", .)

## [1] "a" "B" "c"

• Variable: A quantity, quality, or property that you can measure.
• Observation: A set of values that display the relationship between variables. To be an observation, values need to be measured under similar conditions, usually measured on the same observational unit at the same time.
• Value: The state of a variable that you observe when you measure it.

source: Garret Grolemund

1. Each variable is in its own column
2. Each observation is in its own row
3. Each value is in its own cell

The wide format is generally untidy but found in the majority of datasets

head(iris, 3)

##   Sepal.Length Sepal.Width Petal.Length Petal.Width Species
## 1          5.1         3.5          1.4         0.2  setosa
## 2          4.9         3.0          1.4         0.2  setosa
## 3          4.7         3.2          1.3         0.2  setosa

gather

library("tidyr")
iris_melt <- iris %>%
  tibble::rownames_to_column() %>%
  dplyr::tbl_df() %>%
  gather(flower, measure, contains("al"))
iris_melt

## Source: local data frame [600 x 4]
## 
##    rowname Species       flower measure
##      <chr>  <fctr>        <chr>   <dbl>
## 1        1  setosa Sepal.Length     5.1
## 2        2  setosa Sepal.Length     4.9
## 3        3  setosa Sepal.Length     4.7
## 4        4  setosa Sepal.Length     4.6
## 5        5  setosa Sepal.Length     5.0
## 6        6  setosa Sepal.Length     5.4
## 7        7  setosa Sepal.Length     4.6
## 8        8  setosa Sepal.Length     5.0
## 9        9  setosa Sepal.Length     4.4
## 10      10  setosa Sepal.Length     4.9
## ..     ...     ...          ...     ...

spread

iris_melt %>%
  spread(flower, measure)

## Source: local data frame [150 x 6]
## 
##    rowname    Species Petal.Length Petal.Width Sepal.Length Sepal.Width
##      <chr>     <fctr>        <dbl>       <dbl>        <dbl>       <dbl>
## 1        1     setosa          1.4         0.2          5.1         3.5
## 2       10     setosa          1.5         0.1          4.9         3.1
## 3      100 versicolor          4.1         1.3          5.7         2.8
## 4      101  virginica          6.0         2.5          6.3         3.3
## 5      102  virginica          5.1         1.9          5.8         2.7
## 6      103  virginica          5.9         2.1          7.1         3.0
## 7      104  virginica          5.6         1.8          6.3         2.9
## 8      105  virginica          5.8         2.2          6.5         3.0
## 9      106  virginica          6.6         2.1          7.6         3.0
## 10     107  virginica          4.5         1.7          4.9         2.5
## ..     ...        ...          ...         ...          ...         ...

unite

df %>%
  unite(date, c(year, month, day), sep = "-") -> df_unite

separate, use quotes since we are not refering to objects

df_unite %>%
  separate(date, c("year", "month", "day"))

## Source: local data frame [3 x 4]
## 
##    year month   day value
##   <chr> <chr> <chr> <chr>
## 1  2015    11    23  high
## 2  2014     2     1   low
## 3  2014     4    30   low

using ? or help()

?gather

or using the Help tab next to the Packages tab.

Guess column types and give warnings.

library("readr")
soft <- read_tsv("data/GDS5079.soft", skip = 42, na = "null")

## Warning: 2 parsing failures.
##   row      col   expected             actual
## 35557 10344614 an integer !dataset_table_end
## 35557 NA       6 columns  1 columns

Viewer utility

View(soft)

Using Rstudio, right top panel. Select directly your file. This actually uses readr.

library("tidyr")
library("ggplot2")
iris %>%
  gather(flower, measure, 1:4) %>%
  ggplot()+
  geom_boxplot(aes(x = Species, y = measure, fill = flower))

iris %>%
  ggplot(aes(x = Sepal.Length, y = Sepal.Width, colour = Species))+
  geom_point()+
  geom_smooth(method = "lm", se = FALSE)+
  xlab("Length")+
  ylab("Width")+
  ggtitle("Sepal")

iris %>%
  ggplot(aes(x = Sepal.Length, y = Sepal.Width,
             size = Petal.Length / Petal.Width,
             colour = Species))+
  geom_point()+
  scale_size_area("Petal ratio Length / Width")+
  #scale_colour_brewer(palette = 1, type = "qual")+
  scale_colour_manual(values = c("blue", "red", "orange"))+
  xlab("Sepal.Length")+
  ylab("Sepal.Width")

iris %>%
  ggplot(aes(x = Sepal.Length, y = Sepal.Width))+
  geom_point(colour = "Species") # Error

iris %>%
  ggplot(aes(x = Sepal.Length, y = Sepal.Width))+
  geom_point(colour = "red")

iris_melt %>%
  ggplot()+
  geom_histogram(aes(x = measure, fill = flower))

## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

by default, geom_bar counts. For mapping values use stat = "identity"

iris_melt %>%
  ggplot()+
  geom_bar(aes(x = Species, y = measure, fill = flower), stat = "identity")

iris_melt %>%
  ggplot()+
  geom_bar(aes(x = measure, fill = flower))

iris_melt %>%
  ggplot()+
  geom_bar(aes(x = Species, y = measure, fill = flower), position = "fill", stat = "identity")

iris_melt %>%
  ggplot()+
  geom_bar(aes(x = Species, y = measure, fill = flower), position = "dodge", stat = "identity")

iris_melt %>%
  ggplot()+
  geom_tile(aes(x = Species, y = flower, fill = measure))+
  scale_fill_gradient(low = "yellow", high = "red", limits = c(0, 8))

transparency using the alpha parameter

iris_melt %>%
  ggplot()+
  geom_density(aes(x = measure, fill = Species, colour = Species), alpha = 0.6)+
  facet_wrap(~ flower, scale = "free")+
  theme_bw()

transparency using the alpha parameter

iris_melt %>%
  ggplot()+
  geom_density(aes(x = measure, fill = Species, colour = Species), alpha = 0.6)+
  facet_grid(Species ~ flower, scale = "free")+
  theme_bw()

• data structures by Hadley
• R for data science by Hadley & Garrett
• reading data
• tidy data
• plotting
• ggplot2 documentation by Hadley / Winston
• ggplot2 layer by layer by Hadley
• ggplot2 extensions by Daniel Emaasit
• Excellent ressource on R (in French) Introduction to R by Ewen Gallic

• Hadley Wickham
• Garrett Grolemund
• Jenny Bryan
• Ewen Gallic
• David Robinson
• Eric Koncina
• Tony Heurtaux