malexan
Alexander Matrunich
Recently Published
Data Wragling in R
Slides for Tbilisi R Ladies meet up on the 14th of December, 2017.
String manipulation in R with stringr package
Tbilisi R-Ladies group workshop. June 8th, 2017
Who to get into R
Introduction to R. Tbilisi R-Ladies group. May 18th, 2017
Where are missing FCLs?
After conversion HS codes to FCL of 2013 Eurostat trade data we get 30% of records where HS code is not converted to FCL. 26% of all unique FCL codes from mapping table are not found in resultant trade data set. 3% of all records have multiple matches. We have made some checks of HS ranges from original MDB files. All of them do not provide significant findings in dealing with low rate of successful mapping. HS codes from new trade data sets simply are absent in MDB mapping tables.
Migration from HS+ to HS6
Work with extended country-specific HS codes leads to sizable amount of manual input and could not be automated to the acceptable rate. We try to estimate to what extent we can drop extended HS codes and replace them by standard HS codes of length 6.
Intersection of MDB and Tariffline sets of HS-codes
USA 2011, import: There is no HS codes in tariffline data which are absent in MDB-master file.
Comparison of trade data sources
We want to compare ComTrade data from TF_SOURCE and SWS
How to speed up reading of data from SWS
Reading of data from SWS is really slow. For example, getting of dataset with ~200K rows takes about one hour.
Problem is in inefficient R-code in GetData.processNormalizedResult function from faosws package.
It's possible to optimize R-code there and get some speed improvement, but it's recommended to replace it with C++ code.
План анализа культурных практик
Опрос псковичей про их культурное повидло.
