In SWS-Methodology/hsfclmap: Mapping table between country-specific HS codes and FCL codes

options(width = 120) 
knitr::opts_chunk$set(echo = F)
suppressWarnings(suppressPackageStartupMessages(library(dplyr)))
library(stringr)
library(hsfclmap)
library(fclcpcmap)

subdir <- "OrangeBook"
sourcedir <- "tradeR"


source(file.path(Sys.getenv("HOME"), "r_adhoc", "trade_prevalid_testing", "setupconnection.R"))

if(length(lapply(dir(file.path(Sys.getenv("HOME"), "r_adhoc", "privateFAO", subdir, sourcedir), 
                     full.names = T), 
                 source)) == 0) stop("Files for sourcing not found")

data("hsfclmap")

hsfclmap <- hsfclmap %>% 
  mutate(tocode = trailingDigits(fromcode, tocode, digit = 9)) %>% 
  manualCorrections()

To reduce wrong mapping we want to extract HS chapter component from string which contains full HS code.

Firstly we check does length of HS code equal inside of every reporter-flow dataset.

hsfclmap %>% 
  mutate(fromlength = str_length(fromcode)) %>% 
  group_by(area, flow, validyear) %>% 
  select(fromlength) %>% 
  group_by(fromlength, add = T) %>% 
  summarize(n = n()) %>% 
  mutate(diverse = dense_rank(n),
         maxdiv = max(diverse)) %>% 
  ungroup() %>% 
  filter(maxdiv > 1) %>% 
  left_join(hsfclmap %>% 
              select(area, flow, validyear, fromcode) %>% 
              mutate(fromlength = str_length(fromcode)),
            by = c("area", "flow", "validyear", "fromlength")) %>% 
  group_by(area, flow, validyear, fromlength) %>% 
  sample_n(1) %>% 
  as.data.frame()

hsfclmap %>% 
  # select(fromcode, tocode) %>% 
  mutate(hs2from = str_extract(fromcode, "^.{2}"),
         hs2to   = str_extract(tocode, "^.{2}")) %>% 
  # select(hs2from, hs2to) %>% 
  # distinct() %>% 
  filter(as.numeric(hs2from) > as.numeric(hs2to))

hsfclmap %>% 
  filter(area == "CAN",
         fromcode == "0708900000")

hsfclmap %>% 
  filter(area == "CAN",
         fcl == "420")

hsfclmap %>% 
  filter(area == "CAN",
         fcl == "1075")

hsfclmap %>% 
  select(fromcode, tocode, fcl) %>% 
  mutate(hs2from = str_extract(fromcode, "^.{2}"),
         hs2to   = str_extract(tocode, "^.{2}")) %>% 
  select(hs2from, hs2to) %>% 
  distinct() 

hsfclmap %>% 
  mutate(hs2from = str_extract(fromcode, "^.{2}")) %>% 
  filter(hs2from == "00")

hsfclmap %>% 
  mutate(hs2from = str_extract(fromcode, "^.{2}")) %>% 
  select(hs2from) %>% 
  anti_join(data.frame(hs2 = unique(str_extract(getAgriHSCodes(), "^.{2}")),
                       stringsAsFactors = F),
            by = c("hs2from" = "hs2")) %>% 
  group_by(hs2from) %>% 
  summarize(n = n()) %>% 
  arrange(desc(n))

hsfclmap %>% 
  mutate(hs2from = str_extract(fromcode, "^.{2}")) %>% 
  filter(hs2from == "84") %>% 
  sample_n(10)

fclcpcmap %>% 
  filter(fcl == "1300")

SWS-Methodology/hsfclmap documentation built on May 9, 2019, 11:53 a.m.