Transformer des scripts en fonctions

Introduction à l'écriture de fonctions en R

Richie Cotton

Data Evangelist at DataCamp

Modèle de fonction de base

my_fun <- function(arg1, arg2) {
  # Do something
}

La signature

          function(arg1, arg2)

Le corps

                               {
  # Do something
}

library(readr)
test_scores_geography_raw <- read_csv("test_scores_geography.csv")

library(dplyr)
library(lubridate)
test_scores_geography_clean <- test_scores_geography_raw %>% 
  select(person_id, first_name, last_name, test_date, score) %>%
  mutate(test_date = mdy(test_date)) %>%
  filter(!is.na(score))

library(readr)
test_scores_english_raw <- read_csv("test_scores_english.csv")

library(dplyr)
library(lubridate)
test_scores_english_clean <- test_scores_english_raw %>% 
  select(person_id, first_name, last_name, test_date, score) %>%
  mutate(test_date = mdy(test_date)) %>%
  filter(!is.na(score))

library(readr)
test_scores_art_raw <- read_csv("test_scores_art.csv")

library(dplyr)
library(lubridate)
test_scores_art_clean <- test_scores_art_raw %>% 
  select(person_id, first_name, last_name, test_date, score) %>%
  mutate(test_date = mdy(test_date)) %>%
  filter(is.na(score))

library(readr)
test_scores_spanish_raw <- read_csv("test_scores_spanish.csv")

library(dplyr)
library(lubridate)
test_scores_spanish_clean <- test_scores_spanish_raw %>% 
  select(person_id, first_name, last_name, test_date, score) %>%
  mutate(test_date = mdy(test_date)) %>%
  filter(!is.na(score))

Créer un modèle

import_test_scores <- function() {






}

Coller votre script dans le corps

import_test_scores <- function() {
  test_scores_geography_raw <- read_csv("test_scores_geography.csv")

  test_scores_geography_clean <- test_scores_geography_raw %>% 
    select(person_id, first_name, last_name, test_date, score) %>%
    mutate(test_date = mdy(test_date)) %>%
    filter(!is.na(score))
}

Choisir les arguments

import_test_scores <- function(filename) {         # <- only 1 argument
  test_scores_geography_raw <- read_csv("test_scores_geography.csv")

  test_scores_geography_clean <- test_scores_geography_raw %>% 
    select(person_id, first_name, last_name, test_date, score) %>%
    mutate(test_date = mdy(test_date)) %>%
    filter(!is.na(score))
}

Remplacer les valeurs précises par des arguments

import_test_scores <- function(filename) {
  test_scores_geography_raw <- read_csv(filename)  # <- replace specific filename

  test_scores_geography_clean <- raw_data %>% 
    select(person_id, first_name, last_name, test_date, score) %>%
    mutate(test_date = mdy(test_date)) %>%
    filter(!is.na(score))
}

Généraliser les noms de variables

import_test_scores <- function(filename) {
  test_scores_raw <- read_csv(filename)   # <- variable names generalized

  test_scores_clean <- test_scores_raw %>%       # <- variable names generalized
    select(person_id, first_name, last_name, test_date, score) %>%
    mutate(test_date = mdy(test_date)) %>%
    filter(!is.na(score))
}

Retirer l'affectation finale

import_test_scores <- function(filename) {
  test_scores_raw <- read_csv(filename)

  test_scores_raw %>%  # <- remove assignment
    select(person_id, first_name, last_name, test_date, score) %>%
    mutate(test_date = mdy(test_date)) %>%
    filter(!is.na(score))
}

Utiliser votre fonction

test_scores_geography <- import_test_scores("test_scores_geography.csv")
test_scores_english <- import_test_scores("test_scores_english.csv")
test_scores_art <- import_test_scores("test_scores_art.csv")
test_scores_spanish <- import_test_scores("test_scores_spanish.csv")

Arguments de sample()

x : un vecteur de valeurs dans lequel échantillonner.
size : combien de tirages voulez-vous effectuer à partir de x ?
replace : faut-il échantillonner avec remise ou non ?
prob : un vecteur de poids d'échantillonnage pour chaque valeur de x, totalisant un.

Passons à la pratique !

Introduction à l'écriture de fonctions en R