297 soil samples from Peru's Highlands and Rainforest regions were analyzed using portable spectroradiometers covering 350–2500 nm. Loayza, Hildo from the International Potato Center collected this dataset for predicting soil fertility with machine learning. Spectral preprocessing techniques like Savitzky–Golay smoothing and first derivative transformation were applied.
Use Cases
- Predict soil fertility properties based on Vis–NIR reflectance spectra
- Compare soil characteristics between Highland and Rainforest agroecosystems
- Train machine learning algorithms on spectral data with preprocessing techniques
- Validate portable spectroradiometer performance for soil analysis
Strengths
- 297 soil samples provide a substantial sample size for analysis
- Spectral data covers a wide range from 350 to 2500 nm
- Data originates from two distinct agroecological regions in Peru
Limitations
- Column-level documentation is absent; field semantics must be inferred after download
- Row count is unknown, which may limit suitability assessment
Provenance
- Source
- International Potato Center Harvested Dataverse
- Collection Method
- Soil samples analyzed with portable spectroradiometers and machine learning algorithms
- Freshness
- Last updated 2026-05-04 08:10:07; freshness should be verified
- Geography
- Highlands and Rainforest agroecological regions of Peru