Predictive Demand Disruption Signals for Supply Chain Networks
Author(s)
Advisor(s)
Editor(s)
Collections
Supplementary to:
Abstract
Supply chain networks today are complex networks with various actors spread across the globe. They operate in a volatile, uncertain, and disruption-prone environment which requires them to perceive, react, and respond proactively to effectively manage their operations and maintain customer satisfaction. In this paper, we introduce a signaling methodology to generate early warning signals for a time series upon deviation from the normal pattern, serving as complementary information to demand forecasts. The methodology can be used to detect deviations in the demand curves, which will be propagated through the network to enable the decision makers in taking prompt actions to minimize the effect of the deviation, and better position the supply chain in the wake of disruptions. Relying just on demand forecasts for decision-making can be detrimental as occasionally demand forecasts are unable to capture sudden discrete changes (step up or down) or a turning point (e.g., change from a decreasing time trend to an increasing trend) accurately and rapidly. In such situations, demand disruption signals with characteristics that complement the behavior of demand forecasts play an essential role in proactively assessing and preparing to navigate the disruptions. The developed demand disruption signals leverage bias-identification tracking signals on demand forecasts to proactively detect demand disruption potentiality. Through real-world industrial experiments, our model significantly outperforms typical disruption detection models and are able to capture changes in demand patterns.
Sponsor
Date
2024-05
Extent
Resource Type
Text
Resource Subtype
Paper