XGBoost¶

Machine Learning book: The Elements of Statistical Learning https://web.stanford.edu/~hastie/ElemStatLearn/printings/ESLII_print12.pdf#page=380

how it works¶

https://docs.aws.amazon.com/sagemaker/latest/dg/xgboost-HowItWorks.html

XGBoost: forecast the residuals on different levels and add the forecasts together.

Builds a sequence of weak decision trees (models with low individual accuracy) that progressively improve on the predictions of previous trees. Each tree focuses on correcting the errors made by its predecessors.

strength¶

good for tabular data

limits¶

• need to pass trend and seasonality as external regressors

• Tree based models are not known to do well in extrapolation

multioutput¶

https://xgboost.readthedocs.io/en/stable/tutorials/multioutput.html