Enhancement of the use of machine learning in digital soil mapping

Summary

Machine learning utilises statistical models to identify interesting patterns in data to perform specific tasks without explicit human instructions. These models effectively quantify complex relationships between inputs and outputs, often achieving greater accuracy than simpler models. In digital soil mapping, which models and predicts spatial soil information, machine learning is frequently used. However, these models may be less interpretable and may face issues with input data, such as uncertainty and large errors in the soil data. This PhD project addresses these challenges, and others, ensuring optimally produced soil maps through improved machine learning methodology.