Multivariate probabilistic regression with natural gradient boosting
O'Malley, M.; Sykulski, A.M.; Lumpkin, R.; Schuler, A. (2021). Multivariate probabilistic regression with natural gradient boosting. arXiv (Archive) 2106.03823v1
https://arxiv.org/abs/2106.03823
In: arXiv (Archive). Cornell University. , more
|  |
| Available in | Authors |
|
Document type: Preprint
|
| Authors | | Top |
- O'Malley, M.
- Sykulski, A.M.
- Lumpkin, R.
- Schuler, A.
|
|
|
| Abstract |
Many single-target regression problems require estimates of uncertainty along with the point predictions. Probabilistic regression algorithms are well-suited for these tasks. However, the options are much more limited when the prediction target is multivariate and a joint measure of uncertainty is required. For example, in predicting a 2D velocity vector a joint uncertainty would quantify the probability of any vector in the plane, which would be more expressive than two separate uncertainties on the x- and y- components. To enable joint probabilistic regression, we propose a Natural Gradient Boosting (NGBoost) approach based on nonparametrically modeling the conditional parameters of the multivariate predictive distribution. Our method is robust, works out-of-the-box without extensive tuning, is modular with respect to the assumed target distribution, and performs competitively in comparison to existing approaches. We demonstrate these claims in simulation and with a case study predicting two-dimensional oceanographic velocity data. An implementation of our method is available at https://github.com/stanfordmlgroup/ngboost. |
|
