E-commerce application with RFM segmentation and a hybrid recommendation model addressing cold-start and data scarcity.
Standard recommenders struggle with cold-start users and sparse transaction data. RetOPT needed segmentation-aware recommendations that still work when history is limited.
Implemented RFM customer segmentation, then a hybrid model using transaction, product, and customer features — including segment labels — inside a Django e-commerce application with Looker Studio reporting.
Produced a novel segmentation-based hybrid recommender that mitigates cold-start and scarcity issues — aligned with published RFM clustering research.