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

A 15.9 KB DOCX file from figshare by Hyun-Jin Jung, last updated June 1, 2026, under a CC-BY-4.0 license. The study evaluates year-independent predictability of rice grain protein content using machine learning and multi-year field data. It focuses on data structuring and validation strategies, including leave-one-year-out cross-validation to simulate predictions for unseen growing seasons.

Use Cases

• Training machine learning models for year-independent crop quality prediction based on multi-year field data.
• Evaluating the impact of replicate aggregation on predictive robustness for agricultural traits.
• Simulating nitrogen management strategies using anomaly-based analysis of protein content deviations.

Strengths

• Data is structured with agronomic experimental design in mind, using plot-level and replicate-mean datasets.
• Validation employs a stringent leave-one-year-out cross-validation framework to assess generalization.
• The study reports that residual diagnostics indicated no clear systematic bias across protein levels or nitrogen application rates.

Limitations

• The dataset is very small at 15.9 KB, suggesting limited scope or that it may be a summary document rather than raw data.
• Row count and column-level documentation are unknown, which limits suitability assessment.
• The file format is DOCX, which is not a standard data format and may require extraction of tabular information.

Provenance

Source

Hyun-Jin Jung via figshare.

Collection Method

Likely contains results from multi-year field experiments evaluating rice grain protein content.

Freshness

Last updated 2026-06-01 04:15:17.