Extended SPEC: Analysing Loss Functions for Forecasting Sparse Time Series

Joshua Arnold (), Maximilian Moll and Stefan Pickl
Additional contact information
Joshua Arnold: University of the Bundeswehr Munich
Maximilian Moll: University of the Bundeswehr Munich
Stefan Pickl: University of the Bundeswehr Munich

A chapter in Operations Research Proceedings 2024, 2025, pp 113-119 from Springer

Abstract: Abstract In this work, we focus on the problem of loss function design in the context of Intermittent Demand Forecasting (IDF). In particular, this paper aims to resolve two shortcomings of an existing metric that has been introduced in prior work, namely Stock-keeping-oriented Prediction Error Costs (SPEC). We introduce our own loss function based on some of the same core assumptions made for the design of SPEC. More concretely, our loss also distinguishes stock-keeping and opportunity costs. However, contrary to SPEC, our loss is also applicable to time series that exhibit negative values. Furthermore, we define a set of hyperparameters that enable us to introduce nonlinear temporal dependencies in our loss calculation.

Keywords: Machine Learning; Demand Forecasting; Loss Function Design; Intermittent Demand (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:lnopch:978-3-031-92575-7_16

Ordering information: This item can be ordered from
http://www.springer.com/9783031925757

DOI: 10.1007/978-3-031-92575-7_16

Access Statistics for this chapter

More chapters in Lecture Notes in Operations Research from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().