Predictive mapping using spatiotemporal Ensemble ML (R tutorial)

This is a Rmarkdown tutorial that discusses some aspects of spatial and spatiotemporal data and demonstrated how to use ML, specifically Ensemble ML, to train spatiotemporal models and produce time-series of predictions using 3 published data sets: (1) Daily temperatures for Croatia based on meteorological station data, (2) predictions of soil water content in 3D+T (Cookfarm dataset), (3) spatiotemporal distribution of Fagus sylvatica for continental Europe (see complete lecture). Processing is run using R-spatial packages rgdal and terra, while machine learning is implemented using the mlr framework for Ensemble Machine Learning through stacking. The tutorial is available via: https://gitlab.com/geoharmonizer_inea/odse-workshop-2021/. This document contains snapshots of the Rmarkdown tutorial with a backup of R image and all input files, for an up-to-date version of tutorial please refer to GitLab repository.