A hybrid of sequential rules and collaborative filtering for product recommendation

摘要: Customers' purchase behavior may vary over time. Traditional collaborative filtering (CF) methods make recommendations to a target customer based on the of customers whose preferences are similar those customer; however, do not consider how customers' In contrast, sequential rule-based recommendation method analyzes time extract rules in form: previous [email protected]?purchase current period. If customer's history is conditional part rule, then his/her period deemed be consequent rule. Although rule considers sequence time, it does utilize data for To resolve above problems, this work proposes novel hybrid that combines segmentation-based with KNN-CF method. The proposed uses RFM (Recency, Frequency, and Monetary) values cluster into groups values. For each group customers, extracted from sequences recommendations. Meanwhile, provides Then, results two combined final Experiment show outperforms traditional CF methods.