Title: Forecasting (un-)seasonal demand using geostatistics, socio-economic and weather data

Authors: Flora Babongo; Tapio Niemi; Valérie Chavez-Demoulin; Ari-Pekka Hameri; Patrik Appelqvist

Addresses: Faculty of Business and Economics, University of Lausanne, Lausanne, Switzerland ' Faculty of Business and Economics, University of Lausanne, Lausanne, Switzerland ' Faculty of Business and Economics, University of Lausanne, Lausanne, Switzerland ' Faculty of Business and Economics, University of Lausanne, Lausanne, Switzerland ' Faculty of Management, Aalto University, P.O. Box 11000, FI-00076 Aalto, Finland

Abstract: Accurate demand forecasts are essential to supply chain management. We study the spatial demand variation of seasonal and unseasonal sport goods and demonstrate how demand forecast accuracy can be improved by using geostatistics and linking socio-economic and weather data with order line specific supply chain transactions. We found that the socio-economic features impact the demand of both seasonal and unseasonal products and unseasonal products are impacted more. Weather conditions affect only seasonal products. Cross-validation analyses show that using external information improves demand forecasting accuracy by reducing forecasting error up to 48%. The results can be applied both to the operational demand planning process and to the strategy used when making location-based decisions on supply chain actions, for example, deciding locations for new stores or running marketing campaigns.

Keywords: demand forecasting; seasonal products; socio-economic features; weather; geostatistics; kriging; semivariogram.

DOI: 10.1504/IJBFMI.2019.099069

International Journal of Business Forecasting and Marketing Intelligence, 2019 Vol.5 No.1, pp.103 - 124

Received: 19 May 2018
Accepted: 19 Nov 2018
Published online: 12 Apr 2019 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article