Rice Grain Protein Content Predictions Using Multi-Year Field Data and Machine Learning

Multi-year field data evaluates year-independent predictability of rice grain protein content using machine learning. The dataset, authored by Hyun-Jin Jung and last updated in June 2026, focuses on data structuring and validation strategies aligned with agronomic experimental design. Models were trained on plot-level and replicate-mean data and evaluated with leave-one-year-out cross-validation to simulate unseen growing seasons.

Use Cases

• Predicting rice grain protein content based on multi-year agronomic field data.
• Evaluating model generalization for unseen growing seasons using leave-one-year-out cross-validation.
• Supporting adaptive nitrogen management strategies based on anomaly-based analysis of deviations from year-variety baselines.
• Comparing predictive robustness between plot-level and replicate-aggregated datasets.

Strengths

• Dataset is structured to align with agronomic experimental design, focusing on realistic field conditions.
• Validation uses a stringent leave-one-year-out cross-validation framework to assess year-independent prediction.
• Analysis includes anomaly-based evaluation to capture relative variations beyond dominant temporal effects.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Dataset is very small at 14.4 KB, indicating limited scope.

Provenance

Source

figshare

Collection Method

Multi-year field data collected from agronomic experiments.

Freshness

Last updated 2026-06-01 04:15:16; freshness should be verified.