Dirichlet Proportions Model for Hierarchically Coherent Probabilistic Forecasting
Not on the map. Why?
Abstract (arXiv)
Probabilistic, hierarchically coherent forecasting is a key problem in many practical forecasting applications -- the goal is to obtain coherent probabilistic predictions for a large number of time series arranged in a pre-specified tree hierarchy. In this paper, we present an end-to-end deep probabilistic model for hierarchical forecasting that is motivated by a classical top-down strategy. It jointly learns the distribution of the root time series, and the (dirichlet) proportions according to which each parent time-series is split among its children at any point in time. The resulting forecasts are naturally coherent, and provide probabilistic predictions over all time series in the hierarchy. We experiment on several public datasets and demonstrate significant improvements of up to 26% on most datasets compared to state-of-the-art baselines. Finally, we also provide theoretical justification for the superiority of our top-down approach compared to the more traditional bottom-up modeling.
Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 18 of its 64 references that are in the map. First posted on arXiv 21 Apr 2022.
Builds on
18 of 64 references.
References · 18
- 2017Attention Is All You Needcited by 60,361
- 2022FEDformer: Frequency Enhanced Decomposed Transformer for Long-term Series Forecastingcited by 1,210
- 2017Spatio-Temporal Graph Convolutional Networks: A Deep Learning Framework for Traffic Forecastingcited by 1,136
- 2017DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networkscited by 1,013
- 2019N-BEATS: Neural basis expansion analysis for interpretable time series forecastingcited by 753
- 2020Adaptive Graph Convolutional Recurrent Network for Traffic Forecastingcited by 530
- 2021Spectral Temporal Graph Neural Network for Multivariate Time-series Forecastingcited by 299
- 2019Think Globally, Act Locally: A Deep Neural Network Approach to High-Dimensional Time Series Forecastingcited by 161
- 2019High-Dimensional Multivariate Forecasting with Low-Rank Gaussian Copula Processescited by 148
- 2020Multivariate Probabilistic Time Series Forecasting via Conditioned Normalizing Flowscited by 144
- 2021Neural basis expansion analysis with exogenous variables: Forecasting electricity prices with NBEATSxcited by 64
- 2020A machine learning approach for forecasting hierarchical time seriescited by 29
6 more
- 2021Simultaneously Reconciled Quantile Forecasting of Hierarchically Related Time Seriescited by 24
- 2021Probabilistic Hierarchical Forecasting with Deep Poisson Mixturescited by 16
- 2021Hierarchically Regularized Deep Forecastingcited by 9
- 2021On the benefits of maximum likelihood estimation for Regression and Forecastingcited by 8
- 2019A Self-supervised Approach to Hierarchical Forecasting with Applications to Groupwise Synthetic Controlscited by 3
- 2019Machine learning applications in time series hierarchical forecastingcited by 2
Built on by
Citing papers per year: 2023: 2; 2024: 0; 2025: 1.
3 papers.