Table 1_Year-independent prediction of rice grain protein content using machine learning w

A dataset supporting a study on predicting rice grain protein content under variable field conditions using machine learning. The data includes multi-year field observations structured at plot and replicate levels, used to train models evaluated with leave-one-year-out cross-validation. The dataset was authored by Hyun-Jin Jung and last updated on June 1, 2026.

Use Cases

• Training year-independent prediction models for rice grain protein content based on multi-year agronomic field data.
• Evaluating model generalization using leave-one-year-out cross-validation strategies as described.
• Analyzing the effect of replicate aggregation on reducing plot-level variability for predictive robustness.
• Investigating anomaly-based predictions of deviations from year-variety baselines for crop quality.

Strengths

• Dataset is structured with both plot-level and replicate-mean versions, allowing analysis of aggregation effects.
• Models were validated using a stringent leave-one-year-out framework to simulate unseen growing seasons.
• The study framework is explicitly designed for transferable year-independent prediction in agronomy.

Limitations

• The dataset is very small at 14.9 KB, indicating limited scope.
• Row count and column-level documentation are unknown, requiring manual inspection after download.
• Data may reflect geographic or temporal bias inherent to the specific field trials conducted.

Provenance

Source

Hyun-Jin Jung via figshare.

Collection Method

Collected from multi-year agronomic field experiments.

Freshness

Last updated 2026-06-01 04:15:17; freshness should be verified.