With the explosive growth of online shopping demand, conventional offline-based retailers have started to enter the online shopping market evolving from single channel to omnichannel retailers. In an attempt to leverage existing physical store infrastructure, some of the retailers adopted the in-store fulfillment strategy that utilizes store inventory to fulfill online shopping orders. Then it is crucial to determine which stores should be equipped for online order fulfillment and which product assortment should be planned for each of those stores.
The objective of this thesis is to provide a systematic store location and product allocation strategy for in-store online order fulfillment in an omnichannel retail environment. The existing literature on omnichannel retailing lacks in addressing the specific issues for store location and product allocation strategies based on the actual customer order data. The originality of this thesis lies in its data-driven approach to address such issues by combining spatial cluster detection and conventional association rule mining for identifying location-specific product sales patterns.
A data-driven approach is proposed, which integrates both market basket analysis and spatial cluster detection methods. Hotspot analysis is performed to detect spatial clusters of high-demand locations for the specific product groups identified in the first stage. Second, potential in-store fulfillment service areas are identified based on the driving time from the store to the surrounding districts. Then, the accessibility of each store is calculated based on the online sales volume and driving time to choose the potential stores that are suitable for fulfilling online orders. Finally, stores desirable for in-store fulfillment service are identified, and store-specific product groups are suggested based on association rule mining for customer orders within the service area of each store.