Skip to contents

version 4.5

  • Reflects changes in the API and therefore reads JSON data.
  • Continues to use data.table for the internals and tibble to display the results nicely.
  • Future versions will only use tibble but that would require effort.

version 4.2

Updates * Adds dyadic distances table (ots_distances) for gravity modelling

Breaking changes * The API now returns Parquet files instead of JSON for all the cases (except for available years and tables) * All the tests were adapted to reflect changes in the DB (i.e., the test “multiple country match = error” no longer returns an error because there are no two codes for Germany in 2002-2020) * The command use_localhost was removed because the Shiny app now connects directly to PostgreSQL (i.e., this saves resources) * The column EU-28 member was removed from the countries table (i.e., avoid UK confusion after Brexit)

version 4.1

Updates * Uses GDP deflator to convert dollars from one year to another (previously it used reported inflation) Breaking changes * New port for local instance (8080 -> 4949)

version 4.0

Updates * Provides ysrpc table to visualize products, replacing yc tables. * The new default option is to download imputed data, as there is no direct API access to raw data. * A new method to correct import/exports mismatches was applied and the datasets now start from 1980 until I test enough the model results for older years. * Drops yr-sections and yr-groups tables * Drops ‘group’ columns in the final data (and replaces it for ‘section’ columns) * Allows to access both raw and imputed data from the API * Allows to search for special codes in the API (i.e., e-490 and other codes) * All these changes have resulted in large speedups with both data downloading and the Open Trade Statistics dashboard

Breaking changes * Most of the tables in the API were renamed, as now I made available an imputation method to remove transportation costs (and correct mismatching flows)

version 3.0.3

Minor fixes * Adds attributes to parquet tables (i.e., it now does the joins to add product name, section color, etc.)

Updates * Uses “_name” instead of “_fullname_english” in final tables colnames

version 3.0.2

CRAN release: 2022-03-01

Updates * Allows to obtain tables in Parquet format from the API, giving a speed-up of ~50% for the final user. * Uses tibble instead of DT to produce lighter vignettes

version 3.0.1

Updates * Adds section colors data for visualization, this is taken from the palette used in

Version 3.0

Updates * Removes all references to tables using communities or short names (both unofficial), reflecting changes in the API * The functionality remains the same, but now the end user functions don’t add a 21-colors palette to the data (i.e. see the data section)


  • Switches from HS92 to HS12 to reflect product changes with less aggregation
  • Drops any data from Harvard (communities and short product names) as these depend on using HS92 4 digits, therefore the color palettes were removed as these depended on the communities table
  • The inflation data was trimmed to a window since the year 2000
  • The commodities data now contains information for +5000 products instead of +1200 as the aggregation level changed in the API
  • Adds RTAs and MFN tariffs for gravity modelling

Version 2.0


  • Uses ISO codes as is (affects Aruba, Roumania, Timor-Leste, Antarctica, Saint Barthelemy, Curacao, Sint Maarten and South Sudan)

Version 1.0

CRAN release: 2020-04-03


  • Reflects API changes with less aggregated data
  • Follows UN COMTRADE notation (i.e. commodity instead of product)
  • Does not impute data before hand, which is better for most of gravity models use cases
  • Provides the data exactly as in the API, returning commodity level data to allow users to do their own aggregation
  • Does not drop reference year with inflation adjustment (
  • Takes max and min available years from the API instead of hardcoded values (

Version 0.4.0


  • Includes yrpc-ga, yrpc-sa, yrc-ga and yr-sa tables reflecting API updates
  • Simplifies end-user functions a bit (i.e. removes include_groups option)
  • Optimizes the code a bit, specially at the joins with tables in the package
  • Fixes codes duplication when both product and group/community match for a search
  • Includes both official and shortened section names

Version 0.3.1


  • Removes yrp_short option reflecting last DB changes

Version 0.3


  • Much improved coverage to detect almost any possible error
  • Fixes case in inflation adjustment when year = reference year

Version 0.2.8


  • Adds caching (in memory or on disk) option (partial contributions from
  • Includes forwards and backwards testing for inflation adjustment
  • Testing for in memory caching

Version 0.2.7

CRAN release: 2019-12-04


  • Adds feedback provided by @danidlsa
  • Now ots_create_tidy_data() has both reporter and partner set to “all” by default

Version 0.2.5

CRAN release: 2019-09-19


  • Added dependency on R >= 3.5.0 because serialized objects in serialize/load version 3 cannot be read in older versions of R
  • Minimal changes in ots_create_tidy_data() to allow multiple countries as arguments, in line with API changes from September 2019

Version 0.2.4

CRAN release: 2019-09-17


  • Removes product_code_length
  • The API was updated with simplified parameters and 2018 data

Version 0.2.3

CRAN release: 2019-08-28


  • Fixtures for testthat evaluation


  • Specific Windows error during check

Version 0.2.2


  • Inflation data
  • Inflation adjustment function
  • Minor changes in vignettes

Version 0.2.1


  • Consistent use of colour vs color, color is used from now
  • Fixed available tables description
  • Adds yrp_short to available tables
  • Adds use_localhost option for our own server or users who want to clone the database locally, therefore avoid having a separate branh for server installation