What is LightGBM and how does it differ from XGBoost?

LightGBM is a gradient boosting framework optimized for speed and large-scale datasets using histogram-based tree growth.

LightGBM improves scalability by using histogram-based binning of continuous features, reducing memory usage and speeding up split finding. It grows trees leaf-wise instead of level-wise, which can improve accuracy but may increase overfitting risk. It also supports GPU acceleration and efficient handling of large sparse datasets.