electricsheepafrica/africa-synth-agriculture-post-harvest-losses-nigeria

⚠️ Synthetic dataset — Parameterized from published SSA literature, not real observations. Not suitable for empirical analysis or policy inference.

Nigeria Agriculture – Post Harvest Losses

Dataset Description

Synthetic Supply Chain & Logistics data for Nigeria agriculture sector.

Category: Supply Chain & Logistics
Rows: 110,000
Format: CSV, Parquet
License: MIT
Synthetic: Yes (generated using reference data from FAO, NBS, NiMet, FMARD)

Dataset Structure

Schema

• id: string
• date: string
• state: string
• value: float
• category: string

Sample Data

| id           | date       | state       |   value | category   |
|:-------------|:-----------|:------------|--------:|:-----------|
| REC-00391669 | 2024-04-18 | Cross River |  117.76 | B          |
| REC-00023477 | 2024-02-09 | Bayelsa     |   80.28 | C          |
| REC-00143335 | 2022-05-15 | Akwa Ibom   |  118.22 | B          |
| REC-00165084 | 2022-09-24 | Ebonyi      |  216.35 | A          |
| REC-00479494 | 2023-08-20 | Akwa Ibom   |  106.83 | A          |

Data Generation Methodology

This dataset was synthetically generated using:

1. Reference Sources:

   • FAO (Food and Agriculture Organization) - crop yields, production data
   • NBS (National Bureau of Statistics, Nigeria) - farm characteristics, surveys
   • NiMet (Nigerian Meteorological Agency) - weather patterns
   • FMARD (Federal Ministry of Agriculture and Rural Development) - extension guides
   • IITA (International Institute of Tropical Agriculture) - agronomic research

2. Domain Constraints:

   • Crop calendars and phenology (planting/harvest windows)
   • Agro-ecological zone characteristics (Sahel, Sudan Savanna, Guinea Savanna, Rainforest)
   • Nigeria-specific realities (smallholder dominance, market dynamics, conflict zones)
   • Statistical distributions matching national agricultural patterns

3. Quality Assurance:

   • Distribution testing (KS test, chi-square)
   • Correlation validation (rainfall-yield, fertilizer-yield, yield-price)
   • Causal consistency (DAG-based generation)
   • Multi-scale coherence (farm → state aggregations)
   • Ethical considerations (representative, unbiased)

See QUALITY_ASSURANCE.md in the repository for full methodology.

Use Cases

• Machine Learning: Yield prediction, price forecasting, pest detection, supply chain optimization
• Policy Analysis: Agricultural program evaluation, subsidy impact assessment, food security planning
• Research: Climate-agriculture interactions, market dynamics, technology adoption patterns
• Education: Teaching agricultural economics, data science applications in agriculture

Limitations

• Synthetic data: While grounded in real distributions, individual records are not real observations
• Simplified dynamics: Some complex interactions (e.g., multi-generational pest populations) are simplified
• Temporal scope: Covers 2022-2025; may not reflect longer-term trends or future climate scenarios
• Spatial resolution: State/LGA level; does not capture micro-level heterogeneity within localities

Citation

If you use this dataset, please cite:

@dataset{nigeria_agriculture_2025,
  title = {Nigeria Agriculture – Post Harvest Losses},
  author = {Electric Sheep Africa},
  year = {2025},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/electricsheepafrica/nigerian_agriculture_post_harvest_losses}
}

Related Datasets

This dataset is part of the Nigeria Agriculture & Food Systems collection:

• https://huggingface.co/collections/electricsheepafrica/nigeria-agriculture-and-food-systems

Contact

For questions, feedback, or collaboration:

• Organization: Electric Sheep Africa
• Collection: Nigeria Agriculture & Food Systems
• Repository: https://github.com/electricsheepafrica/nigerian-datasets

Changelog

Version 1.0.0 (October 2025)

• Initial release
• 110,000 synthetic records
• Quality-assured using FAO/NBS/NiMet reference data