Multivariate Boosted Trees and Applications to Forecasting and Control

Nespoli, Lorenzo and Medici, Vasco (2022) Multivariate Boosted Trees and Applications to Forecasting and Control. Journal of Machine Learning Research. ISSN 1533-7928

[img] Text

Download (1MB)


Gradient boosted trees are competition-winning, general-purpose, non-parametric regressors, which exploit sequential model fitting and gradient descent to minimize a specific loss function. The most popular implementations are tailored to univariate regression and classification tasks, precluding the possibility of capturing multivariate target crosscorrelations and applying structured penalties to the predictions. In this paper, we present a computationally efficient algorithm for fitting multivariate boosted trees. We show that multivariate trees can outperform their univariate counterpart when the predictions are correlated. Furthermore, the algorithm allows to arbitrarily regularize the predictions, so that properties like smoothness, consistency and functional relations can be enforced. We present applications and numerical results related to forecasting and control.

Actions (login required)

View Item View Item