Skip to content

View or edit on GitHub

This page is synchronized from trase/models/brazil/beef/qa_beef/zdcs_v2_2_0_versus_v2_2_1.md. Last modified on 2026-03-21 22:30 CET by Trase Admin. Please view or edit the original file there; changes should be reflected here after a midnight build (CET time), or manually triggering it with a GitHub action (link).

Brazil Beef ZDCs - change from v2.2.0 to v2.2.1

FloatProgress(value=0.0, layout=Layout(width='auto'), style=ProgressStyle(bar_color='black'))

Read s3://trase-storage/brazil/beef/sei_pcs/v2.2.1/post_embedding/brazil_beef_sei_pcs_v2_2_1_for_ingest_2010_2023.parquet
Read supply_chains_datasets.brazil_beef_v2_2_0 from the PostgreSQL database

Do we have the same total volume in both datasets?

We simplify the ZDC categories by grouping any commitment (TAC, G4, COMPANY COMMITMENT, G4 & TAC etc.) under simply “COMMITMENT”. The category “NONE” in the older dataset is renamed to “NO COMMITMENT” to match the newer one.

Changes in simplified ZDC categories over time

By plotting our “simplified” ZDC categories we see three things:

  1. Total volume attached to commitment is very slightly higher in v2.2.1
  2. Total volume where the commitment is unknown is much higher in v2.2.1, two- or three-fold
  3. Total volume with a confirmed “no commitment” is much lower in v2.2.1, about half or a third

Focus on 2020

To better understand the picture, we remove the time dimension and focus just on 2020; the last year for which old and new data are both available. We see:

  1. A threefold increase in UNKNOWNs
  2. A halving of NO COMMITMENT
  3. Slightly increased commitments

Our hypothesis is twofold:

  1. The increase in unknowns and decrease in “no commitment” is due to changes in how the unknown biome is handled.
  2. We have slightly more commitments due to improved TAC data.

To test the first hypothesis, let’s slice the data by biome unknown vs known. We also provide the full breakdown by commitment type.

We see that:

  1. For known biomes, the levels of UNKNOWNS are quite similar. It is only in the UNKNOWN biome that we see a big increase in UNKNOWN ZDCs going to v2.2.1.
  2. For the unknown biome, the older dataset had a lot of NO COMMITMENT. The v2.2.1 dataset however, only has UNKNOWN and some COMPANY COMMITMENT in the unknown biome.

Trend for top traders

Finally, let’s look at the trend for the top traders, as indentified by their 2023 volume.

JBS

MARFRIG

MINERVA

NATURAFRIG ALIMENTOS

MATABOI / PRIMA FOODS

FRIGOL

PLENA ALIMENTOS

BARRA MANSA COMERCIO DE CARNES E DERIVADOS

AGRA AGROINDUSTRIAL DE ALIMENTOS