Colombian Customs Microdata
Economics
Numeric
Location
Network
The “final form” of this dataset evolved over a few stages:
- First, Jeff announced in class that the dataset would comprise UN Comtrade data merged with vessel arrival and departure data from IMF PortWatch. Unfortunately, Georgetown does not subscribe to Comtrade, so I’m unable to bulk-export that data
- Next, we tried the UN Food and Agriculture Organization’s FAOSTAT “Detailed Trade Matrix”: this dataset is available in full, but is aggregated to the (sender, receiver, product, year) level
- For the microdata we really want – so that we can simulate the kind of individual financial transaction data that you might encounter as a data scientist in Retail Sales, it turns out that Colombia’s Departamento Administrativo Nacional de Estadística (DANE) provides individual-transaction level data!
With that note on the trajectory of this dataset out of the way, here is your vignette and accompanying sample midterm questions:
Midterm Vignette
Here, for ease of use, we have translated the (abbreviated) column headers into simpler-to-understand English versions. The codebook is as follows:
| Column Name | Translation | Description |
|---|---|---|
FECH |
date |
FECH is short for “fecha”, meaning “date”. Here, the date has four digits, where the first two represent the year and the second two represent the month: YYMM. |
ADUA |
department |
ADUA is short for “Aduana”, meaning “customs house”. Here, the specific sub-department of the customs declaration is encoded as a number. Department 3, for example, processed over 10K exports in January 2025, whereas department 27 processed only 1. |