cleaning solutions: combine case_when and stringer functions in different messy data scenarios

combo-case_when-stringer.R

``` r
if(!require("tidyverse")) install.packages("tidyverse")
if(!require("snakecase")) install.packages("snakecase")
if(!require("janitor")) install.packages("janitor")

library(tidyverse)

# read raw data -----------------------------------------------------------

# scenario:
# - pre-processed in excel
# - manually filled cells

raw_data <- tribble(
  ~id_person, ~test_result, ~food_place, ~birth_place,
  7177,"positiv","mercado belen","Ánchash",
  0177,"positivo","mer. belen","huánuco",
  7178,"pos","en Mercado Belen",NA_character_
)
raw_data
#> # A tibble: 3 x 4
#>   id_person test_result food_place       birth_place
#>       <dbl> <chr>       <chr>            <chr>      
#> 1      7177 positiv     mercado belen    Ánchash    
#> 2       177 positivo    mer. belen       huánuco    
#> 3      7178 pos         en Mercado Belen <NA>

# clean test_result -------------------------------------------------------

# combo:
# - case_when + str_detect

raw_data %>% 
  select(test_result) %>% 
  mutate(test_result_clean = case_when(
    str_detect(test_result,"pos") ~ "positive",
    TRUE ~ test_result
  ))
#> # A tibble: 3 x 2
#>   test_result test_result_clean
#>   <chr>       <chr>            
#> 1 positiv     positive         
#> 2 positivo    positive         
#> 3 pos         positive

# clean food_place --------------------------------------------------------

# combo:
# - snakecase
# - case_when + str_detect

raw_data %>% 
  select(food_place) %>% 
  mutate(food_place=snakecase::to_snake_case(food_place)) %>% 
  mutate(food_placet_clean = case_when(
    str_detect(food_place,"belen") ~ "market_belen",
    TRUE ~ food_place
  ))
#> # A tibble: 3 x 2
#>   food_place       food_placet_clean
#>   <chr>            <chr>            
#> 1 mercado_belen    market_belen     
#> 2 mer_belen        market_belen     
#> 3 en_mercado_belen market_belen

# clean id_person ---------------------------------------------------------

# combo:
# - as.character
# - case_when + str_length + str_replace + regular expression

raw_data %>% 
  select(id_person) %>% 
  mutate(id_person=as.character(id_person)) %>% 
  mutate(id_person_clean = case_when(
    str_length(id_person)==3 ~ str_replace(id_person,"(.+)","0\\1"),
    TRUE ~ id_person
  ))
#> # A tibble: 3 x 2
#>   id_person id_person_clean
#>   <chr>     <chr>          
#> 1 7177      7177           
#> 2 177       0177           
#> 3 7178      7178


# clean birth_place -------------------------------------------------------

# simulate an scenario with repeated observations
sim_data <- raw_data %>% 
  select(birth_place) %>%
  bind_rows(raw_data %>% 
              select(birth_place))

# issue:
# - is creates trailing digits per replicate

sim_data %>%
  mutate(birth_place=janitor::make_clean_names(birth_place))
#> # A tibble: 6 x 1
#>   birth_place
#>   <chr>      
#> 1 anchash    
#> 2 huanuco    
#> 3 na         
#> 4 anchash_2  
#> 5 huanuco_2  
#> 6 na_2

# combo:
# case_when + janitor::make_clean_names + str_replace + regular expression

sim_data %>% 
  mutate(birth_place=case_when(
    !is.na(birth_place)~janitor::make_clean_names(birth_place),
    TRUE ~ NA_character_
  )) %>% 
  mutate(birth_place=str_replace(string = birth_place,
                           pattern = "(.+)_(\\d+)",
                           replacement = "\\1"))
#> # A tibble: 6 x 1
#>   birth_place
#>   <chr>      
#> 1 anchash    
#> 2 huanuco    
#> 3 <NA>       
#> 4 anchash    
#> 5 huanuco    
#> 6 <NA>
```

<sup>Created on 2022-03-09 by the [reprex package](https://reprex.tidyverse.org) (v2.0.1)</sup>