EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

1625. Understand Your Decision Rather than Your Model: Towards Explainable Deep Learning Approaches for Sustainable Commodity Procurement

Invited abstract in session TB-45: Artificial Intelligence and Machine Learning for Decision Support, stream Decision Support Systems.

Tuesday, 10:30-12:00
Room: 30 (building: 324)

Authors (first author is the speaker)

1. Moritz Rettinger
Logistics and Supply Chain Management, TUM School of Management, Technical University of Munich
2. Stefan Minner
TUM School of Management, Technische Universität München

Abstract

Hedging against price increases is particularly important in times of significant market uncertainty and price volatility. For commodity procuring firms, futures contracts are a widespread means of financially hedging price risks. Recently, data-driven decision-support approaches have been developed, with deep learning-based methods achieving particularly good results in capturing non-linear relationships between external features and price trends. We employ a prescriptive deep-learning approach modeling hedging decisions as a multi-label time series classification problem. We backtest the performance in two evaluation periods for natural gas, crude oil, nickel, and copper. The approach performs well in the evaluation period of the testing framework (2018-2020) yet fails to capture the market disruptions (pandemic, geopolitical developments) in the second evaluation period (2021-2023), yielding significant hedging losses or degenerating into a simple hand-to-mouth procurement policy. We employ explainable artificial intelligence to analyze the performance for both periods. The results show that the models fail to perform well under market regime switches and disruptive events within the considered feature set.

Keywords

Status: accepted

Back to the list of papers