correct-20-21-data.R

library(dplyr)
library(tidyr)
library(lubridate)
library(sf)
library(fst)


hk_collisions = hk_collisions %>%
  # Replace "SSPO" district abbr in 2020 and 2021 data to "SSP"
  mutate(district = ifelse(district == "SSPO", "SSP", district)) %>%
  # Set collisions with unknown easting and northing values (`0` in raw data) to NA
  # Also set long and lat to NA
  mutate(
    easting = ifelse(easting == 8e5, NA, easting),
    northing = ifelse(easting == 8e5, NA, northing),
    latitude = ifelse(easting == 8e5, NA, latitude),
    longitude = ifelse(easting == 8e5, NA, longitude),
  )

# Ensure timezone of raw data
lubridate::tz(hk_collisions$date_time) = "Asia/Hong_Kong"



# Get information of all types of vehicles involved in the accidents to show in popup
hk_vehicles_involved <- hk_vehicles %>%
  group_by(serial_no) %>%
  summarize(vehicle_class_involved = paste(sort(unique(vehicle_class)), collapse = ", "))

# Get casualty role involved in each accident to show in popup
casualty_role_n = hk_casualties %>% count(serial_no, casualty_role)

accidents_cas_type <- casualty_role_n %>%
  pivot_wider(
    id_cols = serial_no,
    names_from = casualty_role,
    values_from = n, values_fill = 0
  ) %>%
  rename(cas_ped_n = Pedestrian, cas_pax_n = Passenger, cas_dvr_n = Driver)

# Add date floored to first day of the month for easier month filter handling
collisions_new <- mutate(hk_collisions, year_month = floor_date_to_month(date_time))

hk_collisions_join <- hk_collisions %>%
  left_join(accidents_cas_type, by = "serial_no") %>%
  left_join(hk_vehicles_involved, by = "serial_no") %>%
  # Show full name of district in popup of maps
  left_join(data.frame(DC_Abbr = DISTRICT_ABBR, DC_full_name = DISTRICT_FULL_NAME),
            by = c("district" = "DC_Abbr"))


## ------- Collision Map data ---------

hk_collisions_valid <- filter(hk_collisions_join, !is.na(latitude) & !is.na(longitude))

# Leaflet default expect WGS84 (crs 4326), need custom CRS for HK1980 Grid (crs 2326)
# https://rstudio.github.io/leaflet/projections.html
hk_collisions_valid_sf <- st_as_sf(x = hk_collisions_valid, coords = c("longitude", "latitude"), crs = 4326, remove = FALSE)




write_fst(hk_collisions, "./inst/app/data/hk_collisions.fst")


st_write(hk_collisions_valid_sf, "./inst/app/data/data-manipulated/hk_collisions_valid_sf.gpkg")