Sign in to view source links and access this dataset
Description
Weekly, admin-2 level covariates for Argentina, Brazil, and Mexico from 1979 to 2024, designed for crop yield prediction and benchmarking. The dataset is hosted by author 'notadib' on HuggingFace. A processing pipeline is available on GitHub, which also supports generating data for the USA.
Use Cases
Predict crop yields based on weekly administrative-level covariates.
Benchmark machine learning models for agricultural forecasting.
Analyze long-term trends in crop production factors across Latin America.
Reproduce and extend the dataset for other countries using the provided GitHub scripts.
Strengths
Covers three countries (Argentina, Brazil, Mexico) over a 45-year period (1979–2024).
Provides data at a weekly frequency and admin-2 level granularity.
Includes a reproducible processing pipeline for generating similar data for other regions.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Row count and file size are unknown, which may limit suitability assessment.
The USA data is not included in this release, though it can be reproduced.
Provenance
Source
HuggingFace dataset by author 'notadib'.
Collection Method
Likely compiled from various agricultural and administrative data sources.
Time Range
1979–2024
Freshness
Last updated 2026-05-05 03:13:20.
Geography
Argentina, Brazil, Mexico (admin-2 level)
License is unknown; users should verify licensing terms before use.