LightGBM

Why we included this project

LightGBM is the gradient boosting library most people reach for when they work with tabular data. It builds ensembles of decision trees using a histogram-based approach that keeps training fast and memory use low, and it scales comfortably to large datasets. When a single machine is not enough, it can train in parallel, across distributed clusters, or on GPUs. The project has been around long enough that the Python, R, and command-line interfaces are mature, and the documentation and community answers cover most problems a small team will run into. If your work is mostly structured data rather than deep learning, this is a sensible first thing to try.