fable.intermittent: benchmarking probabilistic forecasting methods for intermittent time series

2026-09-23

AI Summary

This study addresses the fragmentation of probabilistic forecasting methods for intermittent time series and the difficulty of their systematic comparison. Leveraging the R programming language and the fable framework, this work provides a unified implementation of multiple probabilistic forecasting models alongside a standardized evaluation pipeline. The core innovation lies in the proposal of a novel model, TWEES, and the development of an efficient computational package for the Tweedie distribution to overcome intensive computational bottlenecks. Systematic evaluations are conducted across four benchmark datasets. Furthermore, accompanying data packages and acceleration libraries are released concurrently. By unifying model implementations and streamlining computation, this project significantly enhances both forecasting efficiency and model comparability for intermittent time series analysis.

Abstract

Intermittent time series are common in spare-parts demand and retail sales. Since the cost of forecast errors is typically asymmetric, decisions such as inventory control require the full predictive distribution rather than a point forecast. Many probabilistic forecasting methods have been proposed; their implementations, however, are scattered across different software frameworks, making it difficult to compare them systematically. We introduce fable.intermittent, an R package that implements several probabilistic forecasting methods for intermittent series within the fable framework. The package allows several models to be fitted and evaluated on a collection of time series through a single, simple forecasting pipeline. We also introduce TWEES, a new exponential smoothing model with a Tweedie predictive distribution. Fitting TWEES requires repeated evaluation of the computationally demanding Tweedie density. We also release the R package tweedieDistr, whose implementation of the Tweedie distribution is substantially faster than the existing one while preserving the same numerical accuracy. We evaluate the methods implemented in fable.intermittent on four datasets, also released in the package.

Problem

• intermittent time series
• probabilistic forecasting
• benchmarking
• inventory control
• predictive distribution

Innovation

• intermittent time series
• probabilistic forecasting
• Tweedie distribution
• exponential smoothing
• benchmarking