Hierarchical compositional clustering and remote sensing classification of flora in the northern boreal Pallas region, Finland

This dataset is an accompaniment to the manuscript "How to map northern boreal flora: the value of hierarchical clustering and classification".

It consists of:
1. a hierarchical clustering of 488 vegetation survey sites, (`sites.csv`, which includes coordinates in EPSG:3067, and `isopam.RData`, for use with included code);
2. remote sensing (Sentinel-2 and airborne laser scanning) features from the same sites (`rs-data.csv`);
3. random forests to classify remote sensing data to plant community clusters in both hierarchical and flat (non-hierarchical, direct) processes (`hiermodels.RData` and `flatmodels.RData`);
4. confusion matrices evaluating the classifiers (`confmatrix.xlsx`);
5. predicted plant community maps from the study area (directory `predictions`); and
6. R code for producing the clustering, RS features, classification, predictions, as well as figures included in the manuscript and appendices (`main.Rmd` and directory `R`).