'caret' Applications for Spatial-Temporal Models
Description
Supporting functionality to run 'caret' with spatial or spatial-temporal data. 'caret' is a frequently used package for model training and prediction using machine learning. This package includes functions to improve spatial-temporal modelling tasks using 'caret'. It prepares data for Leave-Location-Out and Leave-Time-Out cross-validation which are target-oriented validation strategies for spatial-temporal models. To decrease overfitting and improve model performances, the package implements a forward feature selection that selects suitable predictor variables in view to their contribution to the target-oriented performance.