Paper Number

ECIS2026-2835

Paper Type

SP

Abstract

Accurate sales forecasting during event disruptions is challenging due to dynamic cross-product spillovers and temporal effects like post-promotion dips. Contemporary models often overlook these complex evolving product interdependencies. To this end, we propose a novel Dynamic Sales Forecasting Framework (DSFF) that constructs daily multi-faceted product networks from six behavioral and contextual facets. By integrating these dynamic networks with a temporal self-attention Graph Neural Network (GNN), the DSFF is designed to capture both immediate and lingering sales impacts of events across the product portfolio. This framework provides a systemic and temporal perspective for precise sales forecasting under disruptions, thereby supporting targeted marketing efforts and optimized SKU- level inventory decisions.

Share

COinS
 
Jun 14th, 12:00 AM

Sales Forecasting Under Event Disruption: A Dynamic Product Network Perspective

Accurate sales forecasting during event disruptions is challenging due to dynamic cross-product spillovers and temporal effects like post-promotion dips. Contemporary models often overlook these complex evolving product interdependencies. To this end, we propose a novel Dynamic Sales Forecasting Framework (DSFF) that constructs daily multi-faceted product networks from six behavioral and contextual facets. By integrating these dynamic networks with a temporal self-attention Graph Neural Network (GNN), the DSFF is designed to capture both immediate and lingering sales impacts of events across the product portfolio. This framework provides a systemic and temporal perspective for precise sales forecasting under disruptions, thereby supporting targeted marketing efforts and optimized SKU- level inventory decisions.

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.