Anchoring Bias Detection for Supply Chain Forecasting
Mateus Peixoto
M.Sc. Student, Pontifical Catholic University of Rio de Janeiro
Forecasting demand is a fundamental aspect of SCM, that ensures businesses produce the appropriate type and volume of products, which is vital for sustaining profitability over an extended period. Judgment and forecasting are fundamentally intertwined, the framing of the collected data belongs to the human agent. Despite statistical models being overall more precise judgmental forecasting and adjustment have proved the value of the interference of human agents, which has proved useful from the insertion of context as adjustments to the forecasting models. However, the participation of the human agents also brings the risk of cognitive biases. The most widespread cognitive bias within forecasting is the anchoring bias. The disproportional focus on one specific value, the complexity and context around this bias could not be prevented by simple calculations. Thus, an anchoring bias ontology was created to detect the biases and provide the necessary context for correction, this way integrating statistical modeling and human expertise without the negative effects of the biases, which distort decision-making and lead to undesired consequences. The origins of the biases are not caused by the values themselves but by the context associated with the value and for having a disproportional influence on the agent’s decisions, this also means that biases from multiple sources could manifest at the same time, and for providing an adequate correction an ontology relating the types and context of anchors is paramount, paving the way to automated support for bias detection and correction.
Mateus Peixoto is an MSc candidate in industrial engineering at the Pontifical Catholic University of Rio de Janeiro (PUC-Rio)(2022-), Winner of the 2023 International Institute of Forecasters Student Award (2023). He holds a bachelor's in industrial engineering, with a specialization in the analysis of decision-making and risk. (PUC-Rio 2016-2021). The main research interests are the multilevel integration of supply chains, providing an accurate perception of the potential consequences of decisions, combining of statistical models, robust optimization and automatic mechanisms for bias detection. He has published 3 papers in IPSERA 2024 relating the multi-criteria evaluation of forecasting models, will present a framework for Data-Driven and Cognitive Aware SCM in ENEGEP 2024 and present a new application for Decision-Depedent Uncertainty for scm in the Brazilian symposium of operational research (SBPO).
Development of a Digital Twin for Resilient Multimodal Corridors
Yongjian Tao
EngD candidate, University of Twente.
This talk will present the progress of EngD project focused on building a Digital Twin platform for the Twente Canal. The goal is to enhance resilience in multimodal corridors affected by climate change and other disruptive events such as droughts and floods. The platform integrates real-time monitoring, water level forecasting, and logistics optimization to support decision-making for stakeholders, including logistics companies, government bodies, and industrial users. The discussion will cover the project's current stage, including stakeholder interviews, requirement analysis, and the initial architectural design of the platform.
Yongjian Tao is an EngD candidate at the University of Twente, working on the development of a Digital Twin for multimodal corridors. His current focus is on creating a prototype or demonstrator for the Twente Canal Digital Twin. He hold a Master’s degree in Architecture from Southeast University, China, with a focus on intelligent construction. He has extensive experience in architectural design, digital twin modeling, and urban planning, and have contributed to several digital twin and architectural projects across various industries.