| Summary: | 碩士 === 國立成功大學 === 工業與資訊管理學系碩博士班 === 94 === In a fiercely competitive environment, it is essential that firms can recommend proper products to attract customers and meet their needs. Such recommender systems have now emerged in e-commerce applications to support the recommendation of personal products. Different recommender systems have adopted varied algorithms. Recently, a weighted RFM-based method (WRFM-based method) has been proposed to provide recommendations based on customer lifetime value, including Recency, Frequency, and Monetary. Another algorithm, Preference-based collaborative filtering (CF), typically makes recommendations based on the similarities of customer preferences. The mentioned two algorithms adopt the association rule based recommendation to recommend products. This study proposes a hybrid method that will uncover the merits of the WRFM-based method and the preference-based CF method to improve the quality of recommendations. Experiments are conducted to evaluate the quality of recommendations provided by the proposed method by using a data set concerning the fund marketing. The experimental results indicate that the recommendatory effect of preference outperforms a customer’s lifetime value. The proposed hybrid method outperforms the WRFM-based method and the preference-based CF method. Finally, by analyzing the risk preference of each customer clusters, the proposed hybrid method can indeed provide the basis for personal recommendation in the future.
|
|---|
