# install.packages(c("readxl", "janitor"))
library(tidyverse)
library(readxl)
library(janitor)
d <- read_excel("data-raw/fisheries_data_liberia.xlsx")
d |> glimpse()
# clean names
d <-
  d |> 
  janitor::clean_names()
d |> glimpse()
# how many records per year
d |> count(year)
# the data is by month so lets create a "proper" date that R understands
d <- 
  d |> 
  # this is a two step process:
  #  first we generate something like 2019-01-15" (inside the paste0 stuff
  #  then we convert that to proper R date format using ymd()
  mutate(date = ymd(paste0(year, "-", month_number, "-01")))
# we now have redundant variables (year, month, month_number) so lest get rid of them
d <- 
  d |> 
  select(date, segment, species, 
         latin = scientific_names, # i like shorter variable names
         kg = total_kg,            # again something shorter
         lrd = value_of_prod_lrd)  #   ditto
d |> glimpse()

# lets just do a sum of catches by date and plot it
d |> 
  group_by(date) |> 
  reframe(kg = sum(kg, na.rm = TRUE)) |> 
  ggplot(aes(date, kg)) +
  geom_col() +
  labs(x = "Month", y = "Kg")
# same thing by "segment"
d |> 
  group_by(date, segment) |> 
  reframe(tonnes = sum(kg, na.rm = TRUE) / 1000) |> 
  ggplot(aes(date, tonnes, fill = segment)) +
  geom_col() +
  labs(x = "Month", y = "Tonnes") +
  # nicer colour than the default
  scale_fill_brewer(palette = "Set1")
# The catch by species
catch.by.species <-
  d |> 
  group_by(species) |> 
  reframe(tonnes = sum(kg, na.rm = TRUE) / 1000) |> 
  arrange(desc(tonnes))
catch.by.species |> knitr::kable()
# lump all but top 10 species into "other"
# first get the top ten
top10 <-
  catch.by.species |> 
  slice(1:10) |> # first 10 rows
  pull(species)  # extract species to make a "vector"
top10
# catch by top 10 and then the rest
d |> 
  mutate(species = ifelse(species %in% top10, species, "rest")) |> 
  group_by(date, species, segment) |> 
  reframe(tonnes = sum(kg, na.rm = TRUE) / 1000) |> 
  ggplot(aes(date, tonnes, fill = segment)) +
  geom_col() +
  # separate plot for each species, separate scale on each panel
  facet_wrap(~ species, scales = "free_y") +
  labs(x = "Month", y = "Tonnes") +
  # nicer colour than the default
  scale_fill_brewer(palette = "Set1")