← Back to all news
arXiv Machine Learning

Global tree forecasters collapse at the hierarchical aggregate: a five-panel failure characterization

The paper reports a previously undocumented failure of global gradient‑boosted tree forecasters when applied to hierarchical aggregates. Training a single tree on individual series causes the model to predict a constant outside its training range, leading to severe under‑prediction of the total (30–50× in production and up to 496× in a public M5 reconstruction). The authors characterize this collapse across five datasets, three tree libraries, and multiple training seeds, and demonstrate that simple preprocessing steps—per‑series scaling, weighted aggregate‑level training, or seasonal differencing—can prevent it.

Related stories

arXiv Machine Learning
4d ago

Evaluation Choices Decide the Forecasting Leaderboard: Evidence from a Production Marketplace Panel

The paper demonstrates that the outcome of a forecasting leaderboard is largely determined by the evaluator’s design choices rather than the models themselves. By fixing the data, horizon, and period, the authors varied three key evaluation decisions—unit of analysis, error pooling, and scoring metric—and showed that each can reverse or eliminate the apparent superiority of any forecasting method. The study also evaluates the practical impact of these choices on a deployed system, revealing that the selection rule captures a significant portion of the potential performance gain, and confirms the findings on an external public dataset.

By Md Rezwanul Islam, Wael Mohammed
benchmarks
More like this →
arXiv Machine Learning
Aug 31

Generalized Gibbs Ensemble Weighting for Forecast Combination

The paper introduces Generalized Gibbs Ensemble Weighting (GGEW), a probabilistic framework that assigns weights to forecasting models using a Gibbs-style exponential transformation of normalized predictive loss. GGEW extends basic weighting through numerical stabilization, diversity-aware score corrections, and online hyperparameter adaptation, yielding variants such as Stable Gibbs weighting, Directional Gibbs-NCL, and Symmetric Gibbs-NCL. The authors evaluate GGEW on M4 competition submissions and real-world datasets (Monash Traffic, Electricity, Solar), finding that Gibbs-style adaptive weighting is competitive across various settings, though performance varies by dataset, horizon, and deployment protocol.

By Prasen R. Nuthanakaluva, Nava K. Gaddam
reinforcement-learningbenchmarks
More like this →
Hugging Face Trending Papers
Jun 8

FAME: Forecastability-Aware Mixture of Experts for Heterogeneous Time Series Forecasting

Large-scale retail and industrial forecasting systems contain many heterogeneous time series whose lifecycle, sparsity, volatility, seasonality, spectral patterns, and contextual sensitivity differ substantially. A single forecasting model rarely performs well across all regimes, while dense ensembles increase inference cost and provide limited insight into expert suitability.

efficiencybenchmarks
More like this →
arXiv AI
Jul 14

SciML in the Wild: A Diagnostic Study of When Structural Priors Help and When They Hurt

arXiv:2607. 09684v1 Announce Type: cross Abstract: Scientific Machine Learning (SciML) methods such as Neural Ordinary Differential Equations (NODEs), Physics-Informed Neural Networks (PINNs), and Universal Differential Equations (UDEs) are most effective when structural priors reflect reliable governing dynamics.

By Vrishank Sai Anand, Prathamesh Dinesh Joshi, Raj Abhijit Dandekar, Rajat Dandekar, Sreedath Panat
More like this →
arXiv AI
Aug 17

Forecast Collapse in Time-Series Foundation Models

arXiv:2608. 14106v1 Announce Type: cross Abstract: When forecasting hourly returns for 1,000 US equities, we observe an unexpected phenomenon: predictions become nearly flat and show poor stock ranking, as measured by cross-sectional correlation.

By Shu Wan, Miles Ma, Hank Zhu, Guangqi Liu, Stephen Wang, Qingsong Wen, Huan Liu
benchmarks
More like this →