Foundation Models for Demand Forecasting via Dual-Strategy Ensembling

TL;DR

This paper introduces a dual-strategy ensembling framework that significantly improves foundation models for demand forecasting by capturing hierarchical patterns and integrating diverse model architectures, leading to better accuracy and robustness.

Contribution

It proposes a novel ensemble approach combining hierarchical and architectural strategies to enhance foundation models for real-world sales demand forecasting.

Findings

Outperforms baseline models on M5 and external datasets

Improves accuracy across hierarchical levels

Enhances robustness under distributional shifts

Abstract

Accurate demand forecasting is critical for supply chain optimization, yet remains difficult in practice due to hierarchical complexity, domain shifts, and evolving external factors. While recent foundation models offer strong potential for time series forecasting, they often suffer from architectural rigidity and limited robustness under distributional change. In this paper, we propose a unified ensemble framework that enhances the performance of foundation models for sales forecasting in real-world supply chains. Our method combines two complementary strategies: (1) Hierarchical Ensemble (HE), which partitions training and inference by semantic levels (e.g., store, category, department) to capture localized patterns; and (2) Architectural Ensemble (AE), which integrates predictions from diverse model backbones to mitigate bias and improve stability. We conduct extensive experiments on the M5 benchmark and three external sales datasets, covering both in-domain and zero-shot forecasting. Results show that our approach consistently outperforms strong baselines, improves accuracy across hierarchical levels, and provides a simple yet effective mechanism for boosting generalization in complex forecasting environments.