This might be a silly question but I got confussed. I am using the Tree Cover Density (TCD) raster layer from EEA (https://sdi.eea.europa.eu/catalogue/geoss/eng/catalog.search#/metadata/486f77da-d605-423e-93a9-680760ab6791). The raster has integer values from 0-255, where each integer represents the percentage of tree cover density.
I want to use a machine learning algorithm (XGB) to perform a regression task. XGB doesn't work with categorical layers. I am a bit confused, because although the values of TCD are integers, they don't actually represent classes. My question is: should I use the raster as it is for the regression or I should 'dummify' it (i.e., one hot encoding) before I use it?