In siddbhatia/ships: See most recent routes for a vessel

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(ships)

Download and read raw data

Note: save the data in the inst/extdata folder.

raw_data = read.csv(
  system.file("extdata","ships.csv",package = 'ships',mustWork = TRUE),
  header = T)

Check each ship belongs to only one type

nrow(unique(raw_data[,c("SHIP_ID","ship_type")])) == length(unique(raw_data[,c("SHIP_ID")]))

Get the longest and most recent movement between 2 consecutive records for a ship

library(geosphere)
longest_most_recent <- function(df){
  return(
    df %>%
      mutate(DATETIME = lubridate::as_datetime(DATETIME)) %>%
      arrange(DATETIME) %>%  
      mutate(lag_LON = lag(LON), lag_LAT=lag(LAT), lag_dt = lag(DATETIME)) %>%
      filter(!is.na(lag_LON)) %>%
      rowwise() %>%
      mutate(distance_covered = distm(c(LON, LAT), c(lag_LON, lag_LAT), fun = distHaversine)) %>%
      arrange(desc(distance_covered), desc(DATETIME))  %>%
      head(1)
  )
}
# test longest_most_recent(sample_ship)

Remove ships which have just one observation and apply the function above

single_record_ships <- raw_data %>%
  group_by(SHIP_ID) %>%
  summarise(counts = n()) %>%
  filter(counts<=1) %>%
  select(SHIP_ID) %>%
  unlist() %>%
  unname()

processed_data <-raw_data %>%
  filter(!(SHIP_ID %in% single_record_ships)) %>%
  group_by(SHIP_ID)%>%
  group_split() %>%
  purrr::map_df(~longest_most_recent(.x))

Save data

saveRDS(processed_data,file = "../inst/extdata/longest_most_recent.rds")

siddbhatia/ships documentation built on Dec. 23, 2021, 2:20 a.m.