In this paper, diverse learning tasks such as regression, classification and semi-supervised learning are explained as instances of the same general decision forest model. This unified framework then leads to novel uses of forests, e.g. in density estimation and manifold learning. The corresponding inference algorithm can be implemented and optimized only once, with relatively small changes allowing us to address different tasks. This paper is directed at engineers and PhD students who wish to learn the basics of decision forest s as well as more senior researchers interested in the new research contributions.