Use of min, max, std, and mean in e.g., a RegressorChain to predict e_above_hull and measuring the performance improvement over models trained only on the mean.
conda create -n cbsr python==3.9.*
conda activate cbsr
git clone https://github.com/sparks-baird/composition-based-stability-regression.git
cd composition-based-stability-regression
conda install flit
flit install --pth-file # local installation