SPADE-S: A Sparsity-Robust Foundational Forecaster

TL;DR

SPADE-S is a new forecasting architecture designed to handle heterogeneity, sparsity, and low-magnitude issues in time series data, significantly improving accuracy over existing methods across diverse demand forecasting datasets.

Contribution

The paper introduces SPADE-S, a robust forecasting model that reduces biases related to magnitude and sparsity, outperforming state-of-the-art approaches in various real-world demand forecasting tasks.

Findings

SPADE-S improves forecast accuracy by up to 15%.

Achieves P90 accuracy gains of 2.21%, 6.58%, and 4.28%.

Achieves P50 accuracy gains of 0.92%, 0.77%, and 1.95%.

Abstract

Despite significant advancements in time series forecasting, accurate modeling of time series with strong heterogeneity in magnitude and/or sparsity patterns remains challenging for state-of-the-art deep learning architectures. We identify several factors that lead existing models to systematically underperform on low-magnitude and sparse time series, including loss functions with implicit biases toward high-magnitude series, training-time sampling methods, and limitations of time series encoding methods. SPADE-S is a robust forecasting architecture that significantly reduces magnitude- and sparsity-based systematic biases and improves overall prediction accuracy. Empirical results demonstrate that SPADE-S outperforms existing state-of-the-art approaches across a diverse set of use cases in demand forecasting. In particular, we show that, depending on the quantile forecast and magnitude of the series, SPADE-S can improve forecast accuracy by up to 15%. This results in P90 overall forecast accuracy gains of 2.21%, 6.58%, and 4.28%, and P50 forecast accuracy gains of 0.92%, 0.77%, and 1.95%, respectively, for each of three distinct datasets, ranging from 3 million to 700 million series, from a large online retailer.