Histogram-based decision trees vs depth-wise growth: why LightGBM trains 7x faster on 10M row datasets with equal AUC-ROC scores.
XGBoost traditionally uses level-wise (depth-wise) tree growth, while LightGBM uses leaf-wise tree growth with best-first splitting, significantly accelerating convergence on large datasets.
References & Further Reading:
Core Data Platform Lead. Apache Spark contributor with expertise in Databricks and modern cloud warehouses.