Manipulation de données avec {dplyr} dans le tidyverse

• "deep liar", diplayeur, diplaya
• interdit de dire "déplire"

Aww @hadleywickham wants to change "hadleyverse" to "tidyverse" – consistent, uniform interface so packages work together #useR2016

— Hilary Parker (@hspter) June 29, 2016

Toutes les familles heureuses se ressemblent,
mais chaque famille malheureuse l'est à sa façon.

Like families, tidy datasets are all alike but every messy dataset is messy in its own way

• Dans un jeu de données ordonné, les individus/observations sont en lignes, et les variables en colonnes

age <- c(25, 45, 31, 10)
sexe <- c( "homme", "homme", "femme", "homme")
df <- data.frame(age = age, sexe = sexe)
df

  age  sexe
1  25 homme
2  45 homme
3  31 femme
4  10 homme

"80 % des effets sont le produit de 20 % des causes"

…afin d'être plus opérationnel

• à prononcer "païpe"
• raccourci clavier : "Ctrl + Shift + M"

library(magrittr)

verbe2( 
  verbe1( sujet, complement ),
  ... 
)

devient :

sujet %>% 
  verbe1( complement ) %>%
  verbe2( ... )

head(df)

  age  sexe
1  25 homme
2  45 homme
3  31 femme
4  10 homme

devient :

df %>% head()

  age  sexe
1  25 homme
2  45 homme
3  31 femme
4  10 homme

voire…

df %>% head

  age  sexe
1  25 homme
2  45 homme
3  31 femme
4  10 homme

• quand il n'y a qu'un seul argument, nul besoin d'ajouter les parenthèses

• mettre dans un bol les ingrédients (bowl()) puis,
• ajouter de la farine (append()) puis,
• battre la préparation (beat()) puis,
• former une boule (shape()) puis,
• cuire (bake()) puis,
• refroidir (cool()) et enfin,
• déguster (enjoy()) !

x1 <- bowl(rep("flour", 2), "yeast", "water", "milk", "oil")
x2 <- append(x1, "flour", until = "soft")
x3 <- beat(x2, duration = "3mins")
x4 <- shape(x3, as = "balls", style = "slightly-flat")
x5 <- bake(x4, degrees = 200, duration = "15mins")
x6 <- cool(x5, buns, duration = "5mins")
enjoy(x6)

enjoy(cool(bake(shape(beat(append(bowl(rep("flour", 2), "yeast", "water", "milk", "oil"),
"flour", until = "soft"), duration = "3mins"), as = "balls", style = "slightly-flat"),
degrees = 200, duration = "15mins"), duration = "5mins")) 

• la lecture de la recette se fait de l'intérieur vers l'extérieur

bowl(rep("flour", 2), "yeast", "water", "milk", "oil") %>%     
  append("flour", until = "soft") %>%
  beat(duration = "3mins") %>%
  shape(as = "balls", style = "slightly-flat") %>%
  bake(degrees = 200, duration = "15mins") %>%
  cool(buns, duration = "5mins") %>% 
  enjoy()

• on peut lire %>% comme "ensuite"

• Parce que les data frame c'est has-been…

• Parce que C'est le format choisi par Hadley comme output de ses fonctions read_ de {readr} et {readxl}

• chargement du package :

library(tibble)

• Permet de construire des tableaux

age <- c(25, 45, 31, 10)
sexe <- c( "homme", "homme", "femme", "homme")
df <- data.frame(age = age, sexe = sexe) # sexe devient `factor`
df

  age  sexe
1  25 homme
2  45 homme
3  31 femme
4  10 homme

df <- data.frame(age = age, sexe = sexe, stringsAsFactors = FALSE)
df

  age  sexe
1  25 homme
2  45 homme
3  31 femme
4  10 homme

df_tbl <- tibble(age = age, sexe = sexe)
df_tbl

# A tibble: 4 x 2
    age  sexe
  <dbl> <chr>
1    25 homme
2    45 homme
3    31 femme
4    10 homme

x <- 1:10
names(x) <- paste0("var", 1:10)
x_tbl <- as_tibble(x)
x_tbl

# A tibble: 10 x 1
   value
 * <int>
 1     1
 2     2
 3     3
 4     4
 5     5
 6     6
 7     7
 8     8
 9     9
10    10

x_enframe <- enframe(x, name = "nom", value = "valeur")
x_enframe

# A tibble: 10 x 2
     nom valeur
   <chr>  <int>
 1  var1      1
 2  var2      2
 3  var3      3
 4  var4      4
 5  var5      5
 6  var6      6
 7  var7      7
 8  var8      8
 9  var9      9
10 var10     10

rownames(x_enframe)

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

df_tbl <- tribble( 
  ~age, ~sexe,
  25  , "homme", 
  45  , "homme", 
  34  , "femme", 
  10  , "homme"
)

df_tbl

# A tibble: 4 x 2
    age  sexe
  <dbl> <chr>
1    25 homme
2    45 homme
3    34 femme
4    10 homme

as_tibble(df)

# A tibble: 4 x 2
    age  sexe
  <dbl> <chr>
1    25 homme
2    45 homme
3    31 femme
4    10 homme

• l'affichage raisonné

options(tibble.print_max = 5)
options(tibble.print_min = 2)
df_tbl

# A tibble: 4 x 2
    age  sexe
  <dbl> <chr>
1    25 homme
2    45 homme
3    34 femme
4    10 homme

• le mode de sélection des colonnes

df[ ,"sexe"]

[1] "homme" "homme" "femme" "homme"

df[ ,"sexe", drop = FALSE]

   sexe
1 homme
2 homme
3 femme
4 homme

df_tbl[ , "sexe"]

# A tibble: 4 x 1
   sexe
  <chr>
1 homme
2 homme
3 femme
4 homme

as.data.frame(df_tbl)

  age  sexe
1  25 homme
2  45 homme
3  34 femme
4  10 homme

df_tbl %>% as.data.frame()

  age  sexe
1  25 homme
2  45 homme
3  34 femme
4  10 homme

library(dplyr)
packageVersion("dplyr") # 0.7.1

[1] '0.7.1'

data(starwars)
starwars

# A tibble: 87 x 13
            name height  mass hair_color skin_color eye_color birth_year
           <chr>  <int> <dbl>      <chr>      <chr>     <chr>      <dbl>
1 Luke Skywalker    172    77      blond       fair      blue         19
2          C-3PO    167    75       <NA>       gold    yellow        112
# ... with 85 more rows, and 6 more variables: gender <chr>,
#   homeworld <chr>, species <chr>, films <list>, vehicles <list>,
#   starships <list>

glimpse(starwars)

Observations: 87
Variables: 13
$ name       <chr> "Luke Skywalker", "C-3PO", "R2-D2", "Darth Vader", ...
$ height     <int> 172, 167, 96, 202, 150, 178, 165, 97, 183, 182, 188...
$ mass       <dbl> 77.0, 75.0, 32.0, 136.0, 49.0, 120.0, 75.0, 32.0, 8...
$ hair_color <chr> "blond", NA, NA, "none", "brown", "brown, grey", "b...
$ skin_color <chr> "fair", "gold", "white, blue", "white", "light", "l...
$ eye_color  <chr> "blue", "yellow", "red", "yellow", "brown", "blue",...
$ birth_year <dbl> 19.0, 112.0, 33.0, 41.9, 19.0, 52.0, 47.0, NA, 24.0...
$ gender     <chr> "male", NA, NA, "male", "female", "male", "female",...
$ homeworld  <chr> "Tatooine", "Tatooine", "Naboo", "Tatooine", "Alder...
$ species    <chr> "Human", "Droid", "Droid", "Human", "Human", "Human...
$ films      <list> [<"Revenge of the Sith", "Return of the Jedi", "Th...
$ vehicles   <list> [<"Snowspeeder", "Imperial Speeder Bike">, <>, <>,...
$ starships  <list> [<"X-wing", "Imperial shuttle">, <>, <>, "TIE Adva...

Pour manipuler les observations (lignes) :

• arrange()
• filter()

Pour manipuler les variables (colonnes) :

• select()
• mutate()

• par ordre croissant,
• ou déroissant avec desc()
• et ce en autant de combinaisons de critères de tri

starwars %>% arrange(height)
starwars %>% arrange(desc(height), mass)
starwars %>% arrange(desc(height), mass) %>% tail

starwars %>% arrange(desc(height), mass) %>% tail(3)

# A tibble: 3 x 13
            name height  mass hair_color skin_color eye_color birth_year
           <chr>  <int> <dbl>      <chr>      <chr>     <chr>      <dbl>
1    Poe Dameron     NA    NA      brown      light     brown         NA
2            BB8     NA    NA       none       none     black         NA
3 Captain Phasma     NA    NA    unknown    unknown   unknown         NA
# ... with 6 more variables: gender <chr>, homeworld <chr>, species <chr>,
#   films <list>, vehicles <list>, starships <list>

• les opérations de filtre se font avec les opérateurs logiques/de comparaison : >, <,<=, >=, %in%, &, | , !…

starwars %>% filter(height > 150)
starwars %>% filter(height > 150, height < 200)

starwars %>% filter(height > 150 | species == "Human" )

# A tibble: 75 x 13
            name height  mass hair_color skin_color eye_color birth_year
           <chr>  <int> <dbl>      <chr>      <chr>     <chr>      <dbl>
1 Luke Skywalker    172    77      blond       fair      blue         19
2          C-3PO    167    75       <NA>       gold    yellow        112
# ... with 73 more rows, and 6 more variables: gender <chr>,
#   homeworld <chr>, species <chr>, films <list>, vehicles <list>,
#   starships <list>

starwars %>% filter(height > 150 | species == "Human") %>%
  arrange(desc(mass))

# A tibble: 75 x 13
                   name height  mass hair_color       skin_color
                  <chr>  <int> <dbl>      <chr>            <chr>
1 Jabba Desilijic Tiure    175  1358       <NA> green-tan, brown
2              Grievous    216   159       none     brown, white
# ... with 73 more rows, and 8 more variables: eye_color <chr>,
#   birth_year <dbl>, gender <chr>, homeworld <chr>, species <chr>,
#   films <list>, vehicles <list>, starships <list>

starwars %>% filter(species == "Droid" | species == "Human" |species == "Wookie")

starwars %>% filter(species %in% c("Droid", "Human", "Wookie"))

# A tibble: 40 x 13
            name height  mass hair_color skin_color eye_color birth_year
           <chr>  <int> <dbl>      <chr>      <chr>     <chr>      <dbl>
1 Luke Skywalker    172    77      blond       fair      blue         19
2          C-3PO    167    75       <NA>       gold    yellow        112
# ... with 38 more rows, and 6 more variables: gender <chr>,
#   homeworld <chr>, species <chr>, films <list>, vehicles <list>,
#   starships <list>

starwars %>%
  filter(!is.na(height)) %>%
  arrange(height) %>% tail

# A tibble: 6 x 13
          name height  mass hair_color   skin_color     eye_color
         <chr>  <int> <dbl>      <chr>        <chr>         <chr>
1     Grievous    216   159       none brown, white green, yellow
2 Roos Tarpals    224    82       none         grey        orange
# ... with 4 more rows, and 7 more variables: birth_year <dbl>,
#   gender <chr>, homeworld <chr>, species <chr>, films <list>,
#   vehicles <list>, starships <list>

starwars %>% filter(!is.na(height) & !is.na(mass))

# A tibble: 59 x 13
            name height  mass hair_color skin_color eye_color birth_year
           <chr>  <int> <dbl>      <chr>      <chr>     <chr>      <dbl>
1 Luke Skywalker    172    77      blond       fair      blue         19
2          C-3PO    167    75       <NA>       gold    yellow        112
# ... with 57 more rows, and 6 more variables: gender <chr>,
#   homeworld <chr>, species <chr>, films <list>, vehicles <list>,
#   starships <list>

… dédoublonne un dataset (équivalent de unique())

starwars %>% distinct()

# A tibble: 87 x 13
            name height  mass hair_color skin_color eye_color birth_year
           <chr>  <int> <dbl>      <chr>      <chr>     <chr>      <dbl>
1 Luke Skywalker    172    77      blond       fair      blue         19
2          C-3PO    167    75       <NA>       gold    yellow        112
# ... with 85 more rows, and 6 more variables: gender <chr>,
#   homeworld <chr>, species <chr>, films <list>, vehicles <list>,
#   starships <list>

starwars %>% distinct(hair_color)

# A tibble: 13 x 1
  hair_color
       <chr>
1      blond
2       <NA>
# ... with 11 more rows

starwars %>% distinct(hair_color,.keep_all = TRUE)

# A tibble: 13 x 13
            name height  mass hair_color skin_color eye_color birth_year
           <chr>  <int> <dbl>      <chr>      <chr>     <chr>      <dbl>
1 Luke Skywalker    172    77      blond       fair      blue         19
2          C-3PO    167    75       <NA>       gold    yellow        112
# ... with 11 more rows, and 6 more variables: gender <chr>,
#   homeworld <chr>, species <chr>, films <list>, vehicles <list>,
#   starships <list>

…échantillonnent dans un jeu de données

starwars %>% sample_n(size = 60)

# A tibble: 60 x 13
           name height  mass hair_color   skin_color eye_color birth_year
          <chr>  <int> <dbl>      <chr>        <chr>     <chr>      <dbl>
1 Padmé Amidala    165    45      brown        light     brown         46
2        Ackbar    180    83       none brown mottle    orange         41
# ... with 58 more rows, and 6 more variables: gender <chr>,
#   homeworld <chr>, species <chr>, films <list>, vehicles <list>,
#   starships <list>

starwars %>% sample_n(size = 60) %>% dim

[1] 60 13

starwars %>% sample_frac(size = 0.7)

# A tibble: 61 x 13
          name height  mass hair_color skin_color eye_color birth_year
         <chr>  <int> <dbl>      <chr>      <chr>     <chr>      <dbl>
1 Gregar Typho    185    85      black       dark     brown         NA
2   Mace Windu    188    84       none       dark     brown         72
# ... with 59 more rows, and 6 more variables: gender <chr>,
#   homeworld <chr>, species <chr>, films <list>, vehicles <list>,
#   starships <list>

starwars %>% sample_frac(size = 0.7) %>% dim

[1] 61 13

…sélectionne des observations particulières sur la base de d' indice(s)

starwars %>% slice(25:32)

# A tibble: 8 x 13
    name height  mass hair_color   skin_color eye_color birth_year gender
   <chr>  <int> <dbl>      <chr>        <chr>     <chr>      <dbl>  <chr>
1  Lobot    175    79       none        light      blue         37   male
2 Ackbar    180    83       none brown mottle    orange         41   male
# ... with 6 more rows, and 5 more variables: homeworld <chr>,
#   species <chr>, films <list>, vehicles <list>, starships <list>

starwars %>% slice(which(mass %in% range(mass, na.rm = TRUE)))

# A tibble: 2 x 13
                   name height  mass hair_color       skin_color eye_color
                  <chr>  <int> <dbl>      <chr>            <chr>     <chr>
1 Jabba Desilijic Tiure    175  1358       <NA> green-tan, brown    orange
2         Ratts Tyerell     79    15       none       grey, blue   unknown
# ... with 7 more variables: birth_year <dbl>, gender <chr>,
#   homeworld <chr>, species <chr>, films <list>, vehicles <list>,
#   starships <list>

…trie et filtre les n premières observations

starwars %>% top_n(5, mass)

# A tibble: 5 x 13
                   name height  mass hair_color       skin_color
                  <chr>  <int> <dbl>      <chr>            <chr>
1           Darth Vader    202   136       none            white
2 Jabba Desilijic Tiure    175  1358       <NA> green-tan, brown
3                 IG-88    200   140       none            metal
4              Grievous    216   159       none     brown, white
5               Tarfful    234   136      brown            brown
# ... with 8 more variables: eye_color <chr>, birth_year <dbl>,
#   gender <chr>, homeworld <chr>, species <chr>, films <list>,
#   vehicles <list>, starships <list>

starwars %>% select(name, gender, homeworld, species)

# A tibble: 87 x 4
            name gender homeworld species
           <chr>  <chr>     <chr>   <chr>
1 Luke Skywalker   male  Tatooine   Human
2          C-3PO   <NA>  Tatooine   Droid
# ... with 85 more rows

• -
• starts_with()
• ends_with()
• contains()
• matches()
• one_of()
• :

starwars %>% select(starts_with("h"))

# A tibble: 87 x 3
  height hair_color homeworld
   <int>      <chr>     <chr>
1    172      blond  Tatooine
2    167       <NA>  Tatooine
# ... with 85 more rows

starwars %>% select(matches("._."))

# A tibble: 87 x 4
  hair_color skin_color eye_color birth_year
       <chr>      <chr>     <chr>      <dbl>
1      blond       fair      blue         19
2       <NA>       gold    yellow        112
# ... with 85 more rows

starwars %>% select(hair_color:birth_year)

# A tibble: 87 x 4
  hair_color skin_color eye_color birth_year
       <chr>      <chr>     <chr>      <dbl>
1      blond       fair      blue         19
2       <NA>       gold    yellow        112
# ... with 85 more rows

starwars %>% select(-(films:starships)) # tous sauf les variables listes

# A tibble: 87 x 10
            name height  mass hair_color skin_color eye_color birth_year
           <chr>  <int> <dbl>      <chr>      <chr>     <chr>      <dbl>
1 Luke Skywalker    172    77      blond       fair      blue         19
2          C-3PO    167    75       <NA>       gold    yellow        112
# ... with 85 more rows, and 3 more variables: gender <chr>,
#   homeworld <chr>, species <chr>

starwars %>% select(contains("color"))

# A tibble: 87 x 3
  hair_color skin_color eye_color
       <chr>      <chr>     <chr>
1      blond       fair      blue
2       <NA>       gold    yellow
# ... with 85 more rows

starwars %>% select(one_of("species"))

# A tibble: 87 x 1
  species
    <chr>
1   Human
2   Droid
# ... with 85 more rows

starwars %>% select(one_of("species")) %>% pull()

 [1] "Human"          "Droid"          "Droid"          "Human"         
 [5] "Human"          "Human"          "Human"          "Droid"         
 [9] "Human"          "Human"          "Human"          "Human"         
[13] "Wookiee"        "Human"          "Rodian"         "Hutt"          
[17] "Human"          "Human"          "Yoda's species" "Human"         
[21] "Human"          "Droid"          "Trandoshan"     "Human"         
[25] "Human"          "Mon Calamari"   "Human"          "Human"         
[29] "Ewok"           "Sullustan"      "Human"          "Neimodian"     
[33] "Human"          "Gungan"         "Gungan"         "Gungan"        
[37] NA               "Toydarian"      "Dug"            NA              
[41] "Human"          "Zabrak"         "Twi'lek"        "Twi'lek"       
[45] "Vulptereen"     "Xexto"          "Toong"          "Human"         
[49] "Cerean"         "Nautolan"       "Zabrak"         "Tholothian"    
[53] "Iktotchi"       "Quermian"       "Kel Dor"        "Chagrian"      
[57] "Human"          "Human"          "Human"          "Geonosian"     
[61] "Mirialan"       "Mirialan"       "Human"          "Human"         
[65] "Human"          "Human"          "Clawdite"       "Besalisk"      
[69] "Kaminoan"       "Kaminoan"       "Human"          "Aleena"        
[73] NA               "Skakoan"        "Muun"          
 [ reached getOption("max.print") -- omitted 12 entries ]

Sur la base d'un prédicat :

starwars %>% select_if(is.numeric)

# A tibble: 87 x 3
  height  mass birth_year
   <int> <dbl>      <dbl>
1    172    77         19
2    167    75        112
# ... with 85 more rows

mutate() modifie le jeu de données en largeur pour y adjoindre des colonnes

starwars %>% 
  mutate(imc = mass/(height/100)**2) %>%
  select(name,imc)

# A tibble: 87 x 2
            name      imc
           <chr>    <dbl>
1 Luke Skywalker 26.02758
2          C-3PO 26.89232
# ... with 85 more rows

starwars %>%
  mutate(imc = mass/(height/100)**2,
                    imc = round(imc,1)) %>%
  select(name,imc)

# A tibble: 87 x 2
            name   imc
           <chr> <dbl>
1 Luke Skywalker  26.0
2          C-3PO  26.9
# ... with 85 more rows

Pour éviter d'imbriquer les ifelse():

starwars %>% mutate(imc = mass/(height/100)**2,
                    imc = round(imc,1), 
                    imc_classe = case_when(
                      imc < 16.5 ~ "dénutrition", 
                      imc >= 16.5 & imc < 18.5 ~ "maigreur", 
                      imc >= 18.5 & imc < 25 ~ "normal", 
                      imc >= 25 & imc < 30 ~ "surpoids", 
                      imc >= 30 & imc < 35 ~ "obésité modérée", 
                      imc >= 35 & imc < 40 ~ "obésité sévère", 
                      imc >= 40 ~ "obésité morbide"
                    )) %>% select(name,imc, imc_classe)

# A tibble: 87 x 3
            name   imc imc_classe
           <chr> <dbl>      <chr>
1 Luke Skywalker  26.0   surpoids
2          C-3PO  26.9   surpoids
# ... with 85 more rows

Afin d'appliquer des traitements en masse :

• mutate_all()
• mutate_at()
• mutate_if()

starwars %>% mutate_if(is.numeric, scale)

# A tibble: 87 x 13
            name      height       mass hair_color skin_color eye_color
           <chr>       <dbl>      <dbl>      <chr>      <chr>     <chr>
1 Luke Skywalker -0.06781696 -0.1198643      blond       fair      blue
2          C-3PO -0.21161731 -0.1316667       <NA>       gold    yellow
# ... with 85 more rows, and 7 more variables: birth_year <dbl>,
#   gender <chr>, homeworld <chr>, species <chr>, films <list>,
#   vehicles <list>, starships <list>

starwars %>% mutate_if(is.numeric, funs(scale))

# A tibble: 87 x 13
            name      height       mass hair_color skin_color eye_color
           <chr>       <dbl>      <dbl>      <chr>      <chr>     <chr>
1 Luke Skywalker -0.06781696 -0.1198643      blond       fair      blue
2          C-3PO -0.21161731 -0.1316667       <NA>       gold    yellow
# ... with 85 more rows, and 7 more variables: birth_year <dbl>,
#   gender <chr>, homeworld <chr>, species <chr>, films <list>,
#   vehicles <list>, starships <list>

starwars %>%
  mutate_if(is.numeric, funs(scale = scale)) %>% 
  select(ends_with("scale"))

# A tibble: 87 x 3
  height_scale mass_scale birth_year_scale
         <dbl>      <dbl>            <dbl>
1  -0.06781696 -0.1198643       -0.4432379
2  -0.21161731 -0.1316667        0.1579589
# ... with 85 more rows

starwars %>%
  mutate_at(c("gender","homeworld","species"), as.factor) %>% 
  select(gender:species)

# A tibble: 87 x 3
  gender homeworld species
  <fctr>    <fctr>  <fctr>
1   male  Tatooine   Human
2     NA  Tatooine   Droid
# ... with 85 more rows

équivaut à :

starwars %>%
  mutate_at(vars(gender:species), as.factor) %>% 
  select(gender:species)

# A tibble: 87 x 3
  gender homeworld species
  <fctr>    <fctr>  <fctr>
1   male  Tatooine   Human
2     NA  Tatooine   Droid
# ... with 85 more rows

• décalages : lag(), lead()
• cumulatives : cumsum(), cumprod()…
• rangs : dense_rank(), min_rank()…
• opérations arithmétiques et logiques : +, -, *, et aussi..>, <,<=, >=
• ifelse(), case_when(), coalesce(), na_if()…

• Paramètres de position : mean(), median()
• Comptages : n() qui ne prends JAMAIS de paramètres
• Paramètres de dispersion : var(), sd(), IQR()
• Rangs : quantile(), min(), max()

starwars %>% summarise(taille_moyenne = mean(!is.na(height)), 
                       taille_variance = var(!is.na(height)), 
                       effectif = n(), 
                       nombre_manquantes = sum(is.na(height)), 
                       effectif_calcul = effectif - nombre_manquantes)

# A tibble: 1 x 5
  taille_moyenne taille_variance effectif nombre_manquantes
           <dbl>           <dbl>    <int>             <int>
1      0.9310345      0.06495589       87                 6
# ... with 1 more variables: effectif_calcul <int>

starwars %>% summarise_if(is.numeric, funs(moyenne = mean(!is.na(.)), 
                                           variance = var(!is.na(.)), 
                                           effectif = n()))

# A tibble: 1 x 9
  height_moyenne mass_moyenne birth_year_moyenne height_variance
           <dbl>        <dbl>              <dbl>           <dbl>
1      0.9310345    0.6781609          0.4942529      0.06495589
# ... with 5 more variables: mass_variance <dbl>,
#   birth_year_variance <dbl>, height_effectif <int>, mass_effectif <int>,
#   birth_year_effectif <int>

starwars %>%
  mutate_if(is.character,as.factor) %>%
  summarise_at(c("gender","species"), nlevels)

# A tibble: 1 x 2
  gender species
   <int>   <int>
1      4      37

Le complément circonstanciel est une fonction, un groupe de mots qui indique les circonstances dans lesquelles se réalise l'action du verbe

… c'est la combinaison de group_by() + summarise()

starwars %>% group_by(species) %>% 
  summarise(masse_moyenne = mean(mass,na.rm=TRUE), 
            taille_moyenne = mean(height,na.rm=TRUE))

# A tibble: 38 x 3
     species masse_moyenne taille_moyenne
       <chr>         <dbl>          <dbl>
 1    Aleena         15.00        79.0000
 2  Besalisk        102.00       198.0000
 3    Cerean         82.00       198.0000
 4  Chagrian           NaN       196.0000
 5  Clawdite         55.00       168.0000
 6     Droid         69.75       140.0000
 7       Dug         40.00       112.0000
 8      Ewok         20.00        88.0000
 9 Geonosian         80.00       183.0000
10    Gungan         74.00       208.6667
# ... with 28 more rows

starwars %>% group_by(species) %>% 
  summarise(masse_moyenne = mean(!is.na(mass)), 
            taille_moyenne = mean(!is.na(height)), 
            effectif_calcul = n()) %>% 
  filter(effectif_calcul > 1) %>% 
  arrange(desc(effectif_calcul))

# A tibble: 9 x 4
  species masse_moyenne taille_moyenne effectif_calcul
    <chr>         <dbl>          <dbl>           <int>
1   Human     0.6285714      0.8857143              35
2   Droid     0.8000000      0.8000000               5
# ... with 7 more rows

starwars %>% group_by(gender) %>% tally()

# A tibble: 5 x 2
         gender     n
          <chr> <int>
1        female    19
2 hermaphrodite     1
3          male    62
4          none     2
5          <NA>     3

équivaut à :

starwars %>% count(gender)

# A tibble: 5 x 2
         gender     n
          <chr> <int>
1        female    19
2 hermaphrodite     1
3          male    62
4          none     2
5          <NA>     3

• bind_cols() pour remplacer cbind()
• left_join(), right_join(), inner_join(), full_join(), anti_join()…

apprentisage <- starwars %>% select(-(films:starships)) %>% sample_frac(0.7)
validation <- starwars %>% select(-(films:starships))%>% anti_join(apprentisage)
dim(validation)

[1] 26 10

dim(apprentisage)

[1] 61 10

• bind_rows() pour remplacer rbind() : nous épargne les galères de facteurs
• intersect(x,y) : observations présentes en x et en y
• setdiff(x,y) : observations présentes en x et absentes en y
• union() : observations présentes en x ou y (union_all() conserves les doublons)

La recette pour nettoyer ses données :

library(tidyr)

experience <- data.frame(id = c(1,2,3,4),
                         sexe = c("F","M","M","F"),
                         contrôle = c(7.9,6.3,9.5,11.5),
                         traitement = c(12.3,10.6,13.1,13.4))
experience %>% 
gather(key = condition, value = glycémie,contrôle:traitement)

  id sexe  condition glycémie
1  1    F   contrôle      7.9
2  2    M   contrôle      6.3
3  3    M   contrôle      9.5
4  4    F   contrôle     11.5
5  1    F traitement     12.3
6  2    M traitement     10.6
7  3    M traitement     13.1
8  4    F traitement     13.4

temp_europ <- data.frame(
Ville = c(rep("Amsterdam",3),rep("Lisbonne",3)), 
Mois = c(rep(c("Janvier", "Février", "Mars"),2)), 
Température = c(2.9,2.5,5.7, 10.5,11.3, 12.8))

temp_europ

      Ville    Mois Température
1 Amsterdam Janvier         2.9
2 Amsterdam Février         2.5
3 Amsterdam    Mars         5.7
4  Lisbonne Janvier        10.5
5  Lisbonne Février        11.3
6  Lisbonne    Mars        12.8

temp_europ %>% 
spread(key = Mois, value = Température)

      Ville Février Janvier Mars
1 Amsterdam     2.5     2.9  5.7
2  Lisbonne    11.3    10.5 12.8

starwars %>% group_by(gender, species) %>% 
  tally() %>%
  spread(key = species, value = n, fill = 0)

# A tibble: 5 x 39
# Groups:   gender [5]
         gender Aleena Besalisk Cerean Chagrian Clawdite Droid   Dug  Ewok
*         <chr>  <dbl>    <dbl>  <dbl>    <dbl>    <dbl> <dbl> <dbl> <dbl>
1        female      0        0      0        0        1     0     0     0
2 hermaphrodite      0        0      0        0        0     0     0     0
3          male      1        1      1        1        0     0     1     1
4          none      0        0      0        0        0     2     0     0
5          <NA>      0        0      0        0        0     3     0     0
# ... with 30 more variables: Geonosian <dbl>, Gungan <dbl>, Human <dbl>,
#   Hutt <dbl>, Iktotchi <dbl>, Kaleesh <dbl>, Kaminoan <dbl>, `Kel
#   Dor` <dbl>, Mirialan <dbl>, `Mon Calamari` <dbl>, Muun <dbl>,
#   Nautolan <dbl>, Neimodian <dbl>, `Pau'an` <dbl>, Quermian <dbl>,
#   Rodian <dbl>, Skakoan <dbl>, Sullustan <dbl>, Tholothian <dbl>,
#   Togruta <dbl>, Toong <dbl>, Toydarian <dbl>, Trandoshan <dbl>,
#   `Twi'lek` <dbl>, Vulptereen <dbl>, Wookiee <dbl>, Xexto <dbl>, `Yoda's
#   species` <dbl>, Zabrak <dbl>, `<NA>` <dbl>

dataset <- data.frame(id = 1:6,annee = c("2014",NA,NA,"2015",NA,NA))

dataset

  id annee
1  1  2014
2  2  <NA>
3  3  <NA>
4  4  2015
5  5  <NA>
6  6  <NA>

dataset %>% fill(annee)

  id annee
1  1  2014
2  2  2014
3  3  2014
4  4  2015
5  5  2015
6  6  2015

dataset <- data.frame(id = 1:6, tension = c("12/8","12/7","14/4","18/10","13/8","12/8"))
dataset

  id tension
1  1    12/8
2  2    12/7
3  3    14/4
4  4   18/10
5  5    13/8
6  6    12/8

dataset %>% 
separate(tension,into = c("PAS","PAD"),sep = "/",remove = TRUE)

  id PAS PAD
1  1  12   8
2  2  12   7
3  3  14   4
4  4  18  10
5  5  13   8
6  6  12   8

dataset <- data.frame(id = 1:6,
    annee = c("2012","2012","2013","2014","2014","2014"),
    mois = month.abb[c(1,3,2,1,2,3)],
    temperature = c(12,11,15,16,18,10))
dataset

  id annee mois temperature
1  1  2012  Jan          12
2  2  2012  Mar          11
3  3  2013  Feb          15
4  4  2014  Jan          16
5  5  2014  Feb          18
6  6  2014  Mar          10

dataset %>% complete(annee, mois)

# A tibble: 9 x 4
   annee   mois    id temperature
  <fctr> <fctr> <int>       <dbl>
1   2012    Feb    NA          NA
2   2012    Jan     1          12
3   2012    Mar     2          11
4   2013    Feb     3          15
5   2013    Jan    NA          NA
6   2013    Mar    NA          NA
7   2014    Feb     5          18
8   2014    Jan     4          16
9   2014    Mar     6          10

Autres packages utiles du tidyverse :

• {readr} : l'import réussi des données
• {purr} : programmation fonctionnelle
• {lubridate} : manipulation des dates
• {forcats} : manipulation des facteurs
• {roxygen2} et devtools : mettre le tout en package

• la cheatsheet "Data Transformation" de RStudio
• la cheatsheet "Data Import" de RStudio