# The pacman package has the p_load() function that can check if a package is installed and if not it will install it.
# The DataExplorer package has the create_report() function that uses AutoEDA to product a report about all variables.

# https://cran.r-project.org/web/packages/DataExplorer/vignettes/dataexplorer-intro.html#slicing-dicing


library(pacman)  

p_load(nycflights13, tidyverse, visdat, naniar, DataExplorer)  

flights

airlines
planes
airports

weather

data_list <- list(airlines, airports, flights, planes, weather)
plot_str(data_list)

airlines <- airlines %>% rename(name_airlines = name)
head(airlines)
merge_airlines <- left_join(flights, airlines, by = "carrier")
head(merge_airlines)
planes <- planes %>% rename(year_manufacture = year)
head(planes)
merge_planes <- left_join(merge_airlines, planes, by = "tailnum")
head(merge_planes)
airports <- airports %>% rename(name_airport = name)
head(airports)
merge_airports_origin <- left_join(merge_planes, airports, by = c("origin" = "faa"))
head(merge_airports_origin)

final_nycflights13 <- merge_airports_origin

final_nycflights13 %>% introduce()
final_nycflights13 %>% plot_intro()
final_nycflights13 %>% plot_missing()

final_nycflights13 <- final_nycflights13 %>% select(-speed)

final_nycflights13 %>% introduce()
final_nycflights13 %>% plot_intro()
final_nycflights13 %>% plot_missing()

final_nycflights13 %>% plot_bar()

final_nycflights13 <- final_nycflights13 %>% mutate( manufacturer = fct_collapse(manufacturer, 
                               AIRBUS = c("AIRBUS INDUSTRIE", "AIRBUS"), 
                               "MCDONNELL DOUGLAS" = c("MCDONNELL DOUGLAS AIRCRAFT CO", "MCDONNELL DOUGLAS CORPORATION","MCDONNELL DOUGLAS") )) 

final_nycflights13 %>% select(manufacturer) %>%
  plot_bar()

final_nycflights13 %>% introduce()

final_nycflights13 <- final_nycflights13 %>% select(-dst, -tzone, -tz)

final_nycflights13 %>% introduce()

final_nycflights13 %>% plot_bar()

# conditional on arr_delay

final_nycflights13 %>% plot_bar(with = "arr_delay")

final_nycflights13 %>% plot_histogram()

plot_correlation(na.omit(final_nycflights13), maxcat = 5L)

plot_correlation(na.omit(final_nycflights13), type = "c")
plot_correlation(na.omit(final_nycflights13), type = "d")


config <- configure_report(
  add_plot_str = FALSE,
  add_plot_qq = FALSE,
  add_plot_prcomp = FALSE,
  add_plot_boxplot = FALSE,
  add_plot_scatterplot = FALSE,
  global_ggtheme = quote(theme_minimal(base_size = 14))
)
create_report(final_nycflights13, y = "arr_delay", config = config)