Calculating the prediction interval coverage probability (PICP)

In digital soil mapping (DSM), we make predictions of the spatial distribution of a soil property, which comes with uncertainties/errors. To quantify the accuracy we split the data into a training and test set, where we train a machine learning model (e.g., random forest, additive models, splines, etc.) ...

accuracy data digital distribution errors etc interval machine machine learning machine learning model mapping prediction predictions probability property random r bloggers reading set test training