M5

The M5 dataset is a large-scale real-world retail dataset used to evaluate time series forecasting models, capturing complex dynamics across multiple variables.

Papers using M5 (12)

Foundation Models for Demand Forecasting via Dual-Strategy Ensembling2025 · 3 citesAME-TS: Anchored Mixture-of-Experts for Time Series Forecasting2026Time-Aware Prior Fitted Networks for Zero-Shot Forecasting with Exogenous Variables2026Let Experts Feel Uncertainty: A Multi-Expert Label Distribution Approach to Probabilistic Time Series Forecasting2026Faithful and Interpretable Explanations for Complex Ensemble Time Series Forecasts using Surrogate Models and Forecastability Analysis2025Hierarchical Time Series Forecasting Via Latent Mean Encoding2025Deep Learning for Time Series Forecasting: Tutorial and Literature Survey2020 · 345 citesTSMixer: An All-MLP Architecture for Time Series Forecasting2023 · 124 citesHierarchical Proxy Modeling for Improved HPO in Time Series Forecasting2022 · 2 citesPastprop-RNN: improved predictions of the future by correcting the past2021Hierarchical Forecasting at Scale2023Scalable Probabilistic Forecasting in Retail with Gradient Boosted Trees: A Practitioner's Approach2023