Predictive performance of machine learning model with varying sampling designs, sample sizes, and spatial extents

Summary

This 2023 methodological paper examines how experimental design decisions—specifically sampling strategy, sample size, and spatial extent—affect machine learning model performance in environmental and ecological monitoring applications. The work addresses a practical challenge for practitioners designing monitoring programmes by quantifying trade-offs between sampling effort and predictive accuracy. Specific quantitative thresholds and optimal design recommendations would need to be confirmed from the full text.

UK applicability

The methodological framework may be applicable to UK environmental monitoring and soil/agricultural surveillance programmes that employ machine learning for prediction and mapping. Applicability depends on whether the study's spatial contexts and environmental variables align with UK conditions.

Key measures

Predictive performance metrics (likely including accuracy, RMSE, or similar measures) under varying sampling designs, sample sizes, and spatial extents

Outcomes reported

The study evaluated how sampling strategy, sample size, and spatial extent influence the predictive accuracy of machine learning models in environmental monitoring. Quantitative relationships between design parameters and model performance metrics were assessed.

Topic tags