LightGBM (Light Gradient Boosting Machine) is an open-source gradient boosting framework developed by Microsoft and released in 2016. It implements gradient boosting, which improves predictions by sequentially adding decision trees, and is designed for fast training and low memory use on large datasets through histogram-based learning and leaf-wise tree growth.
Along with XGBoost and CatBoost, it is one of the most widely used libraries for classification, regression, and ranking on tabular data, with interfaces for Python, R, C++, and other languages. It is a representative tool of the traditional machine learning approach, in which each task requires its own feature engineering and model training.