Sales Prediction and Product Recommendation Model Through User Behavior Analytics

The COVID-19 has brought us unprecedented difficulties and thousands of companies have closed down. general public responded to call the government stay at home. Offline retail stores been severely affected. Therefore, in order transform a traditional offline sales model B2C improve shopping experience, this study aims utilize historical data for exploring, building prediction recommendation models. A novel science life-cycle process with Recency, Frequency, Monetary (RFM) analysis method combination various analytics algorithms are utilized product through user behavior analytics. RFM is segmenting customer levels company identify importance each level. For purchase model, XGBoost Random Forest machine learning used build models 5-fold Cross-Validation evaluate their. association rules theory Apriori algorithm complete basket recommend products according outcomes. Moreover, some suggestions proposed marketing department Overall, achieved better performance accuracy F1-score around 0.789. provides good results combinations improving market responsiveness. Furthermore, it specific new customers. This offered very practical useful business transformation case that assists similar situations their