Chair of Operations, Economics & Strategy

A New Measure for Decision-Making under Uncertainty

Published:13.11.25 — Decision-makers in business, finance, and policy frequently face situations where the best course of action is unclear—not because outcomes are risky in a probabilistic sense, but because essential information is missing. A new study by Prof. Thomas A. Weber (Chair of Operations, Economics and Strategy, EPFL) introduces a general metric to quantify such uncertainty based on observed decisions and outcomes. The paper, titled “A Measure of Decision-Based Payoff Uncertainty,” was presented at the 6th International Conference on Problems of Cybernetics and Informatics (PCI) and published by IEEE.

Professor Weber Wins the 2025 ACIIS Best Paper Award

Published:31.10.25 — At the 2025 IEEE 7th International Conference on Applied Computational Intelligence in Information Systems (ACIIS), held October 20–22 at Universiti Teknologi Brunei in Brunei Darussalam, Prof. Weber received the Best Paper Award for his work on a robust aggregate scoring system for evaluating digital services. The winning paper, “Relatively Robust QoS and QoE Score Aggregation,” is forthcoming on IEEE Xplore as part of the conference proceedings.

Predicting the Status of Project Portfolios Using Markov Chains

Published:24.09.25 — Firms often manage hundreds of concurrent projects with uncertain status trajectories. A new conference paper by Prof. Thomas A. Weber (Chair of Operations, Economics and Strategy, EPFL) introduces a finite-state Markov-chain framework that tracks and forecasts the evolving composition of project portfolios. The method yields forward-looking estimates of success rates, average durations, and resource balance across active and idle projects, supporting earlier risk detection and data-driven reprioritization at the portfolio level. Presented at the 2025 IEEE 5th International Conference on Smart Information Systems and Technologies (SIST), the work demonstrates that the forecasts remain robust even when the underlying identification is moderately noisy, making the approach practical for real organizations with imperfect data.